Findings

2.1 Our findings are summarised into six key areas:

• NISRA’s release and communication of Census 2021 outputs,
• the quality assurance of the census estimates,
• supporting the ongoing use of census data affected by the pandemic,
• meeting user needs and collaboration with other parts of the UK and Republic of Ireland,
• NISRA’s user engagement and,
• means of dissemination.

NISRA’s release and communication of Census 2021 outputs

2.2 NISRA has continued to put users at the centre of its planning and dissemination of the Census 2021 outputs. NISRA conducted two public consultations in 2018 and 2021: one on the proposed strategy for the Census 2021 outputs and another on the outputs’ content and release timeline. NISRA’s publication of the responses has meant that users can understand how their input shaped the final release plans in a transparent way.

2.3 NISRA published the first results from Census 2021 on 24 May 2022. This was around 14 months after the Census was taken (on 21 March 2021), and 2 months sooner than done for the 2011 Census. The release provided headline population statistics for Northern Ireland by age and sex, number of households with usual residents and the communal establishment population. Users are kept up to date with published and upcoming releases by NISRA’s Census 2021 outputs prospectus. Generally, census outputs in NI are grouped into three phases:

• Census 2021 First Release: Population and household statistics. Headline population estimates in NI were published on 24 May 2022. Small area geographies for Data Zones (DZs), Super Data Zones (SDZs), District Electoral Areas (DEAs), unit-level postcodes, postcode sector and districts in Northern Ireland were published in February 2023 and 2024.
• Census 2021 Second Release: Main statistics by topic with NI and Local Government District (LGD) breakdowns were published between September 2022 and March 2023. Supplementary geography breakdowns and main statistics on approximated social grade were published in 2023.
• Census 2021 Third Release: Flexible Table Builder and multivariate data. A selection of multivariate tables were published in May 2023, including religion, national identity and economic activity, ahead of the release of the Flexible Table Builder in June 2023. Further updates to the Flexible Table Builder were made in 2023 and 2024 providing additional geography and topic variables.

2.4 Most Census 2021 outputs were released in line with their pre-announced publication dates; the only exception was a minor delay when all NISRA outputs were paused due to the death of the monarch in September 2022. Any corrections made to census data have been made in line with NISRA’s policy, with any issues also highlighted on the Quality notes, issues and corrections page of the NISRA website. Five corrections have been made to census data spanning different topic areas and, on each occasion, NISRA has provided information on the extent, scale and impact of the correction made. The last correction was made in May 2024 to tables presenting historical trends on marital status to account for changes in categories over time.

2.5 Most users that we spoke to were happy with the timeliness of the NISRA outputs, with a minority sharing frustrations around the time taken to release them. However, the overall user sentiment was positive in relation to NISRA’s delivery of census outputs.

2.6 NISRA effectively promoted its census releases. For example, NISRA announced its first outputs with a suite of communications, including a press release, briefing, tailored presentations to key stakeholders and the use of social media. Users were notified about the releases via email alerts and social media posts. This proactive approach to announcing the outputs continued throughout the release timeline. Overall, the users that we spoke to were very complimentary about NISRA’s communication and engagement.

Quality assurance of the census estimates

Census response rates and use of methods

2.7 Census 2021 involved a move from a traditional paper-based questionnaire to a digital-first data collection approach. Census returns were made either online (just over 80%) or on paper. NISRA achieved a person response rate of 97% in Northern Ireland, the highest response rate of the last three censuses (2001, 2011 and 2021), as highlighted in its Statement about data quality. The move to electronic questionnaires also resulted in higher item level response. NISRA’s analysis of the first five questions on the 2021 questionnaire (first name, last name, date of birth, sex and marital status) showed that the electronic questionnaires had a 99.5% item response rate, while the paper questionnaires had a 96.4% item response rate.

2.8 NISRA worked with ONS and NRS and used international good practice and principles in developing its methods. The Demographics Statistics Advisory Group, consisting of external experts, advised NISRA on how to maximise the relevance, quality and usefulness of statistics produced from the NI census. The group was also a consultative forum providing technical advice and guidance on methodological developments.

2.9 The majority of methods used in 2021 remained largely unchanged from 2011. To address any potential bias in the estimates, NISRA used robust methodologies for data processing and estimation to account for participants who did not respond or had missed particular questions. Those who did not respond were estimated and modelled using two processes (as was done for the 2011 Census). The Census 2021 Census Under Enumeration (CUE) process added 26,900 high quality administrative records into 12,700 non-responding households where evidence pointed to the presence of usual residents at the time of the census. Separately, the Census Coverage Survey (CCS), a subsequent representative sample of households, was then used to estimate any coverage error from the remaining non-responding households, and to estimate those missed from counted households. This led to an adjustment of around 31,900 people and 9,200 households. The CUE adjustment was equivalent to 1.41% of the total population and 1.65% of total households, while the CCS added a further 1.67% and 1.20% respectively.

2.10 NISRA put measures in place to maximise the response rates of special population groups (such as students, homeless people, Irish Travellers), including engagement with relevant organisations and community groups. Given the challenges in counting full-time students observed in the 2019 census rehearsal, NISRA took additional steps to ensure that students were fully counted. Students were encouraged to make returns at their home and term-time addresses. A post-collection process was then used to address any overcount/duplication. A published technical paper details the enumeration and data processing phases for the student population in NI.

Quality Assurance

2.11 NISRA used a rigorous quality assurance approach to understand and assure the reliability of the census results before publication. Guided by its 2021 Census Quality Assurance Strategy, NISRA used a range of quality assurance activities to ensure that the census data were robust and fit for purpose. These efforts included seeking feedback from stakeholder boards, groups and panels comprising a wide range of people and organisations. The expert quality assurance (QA) panel reported directly to the Northern Ireland Registrar General and provided additional assurance. To test the accuracy of the census estimates, NISRA also benchmarked census data against administrative datasets and other statistical releases. NISRA were able to assess the plausibility of the results, for example with the NI population and household estimates, by comparing them to 2021 health, education, pension and electoral register administrative datasets.

2.12 NISRA also carried out a Census Quality Survey (CQS) between October and December 2021 to assess the overall quality of the census. The CQS was a voluntary survey across a random sample of households that had responded to Census 2021. The characteristics of the survey respondents were compared to those of the census, and individual responses were weighted to ensure that the sample was representative (by age, sex, religion). Analysis of the responses was done by matching the individuals (3,936) in the same 1,511 households in the census and comparing agreement rates with confidence intervals to account for sample variability. The results from the Census Quality Survey published in 2023 demonstrated that the Census 2021 is of high quality and fit for purpose.

2.13 The survey asked seven household questions and 33 individual questions. 40% of the CQS questions had an agreement rate above 90% and 67.5% of questions (27 out of 40 questions) had agreement rates over 85%. However, 12 individual questions had an agreement rate lower than 85%. In its analysis, NISRA recognises that the census and CQS responses may not align and explores in more detail the reasons behind the differences. As the CQS was conducted seven months later than the census, it is reasonable to expect that a change in circumstances over time and who in the household provided the response may have led to a difference in responses. Given the timing of the census, an impact on individual responses (such as hours worked) could also be expected due to the pandemic, yet there is no specific mention or acknowledgement of this in NISRA’s analysis.

2.14 The most notable difference is with the ‘Ever worked’ question. The wording of the question was updated from the Census 2011, and the responses indicated an anomaly when compared to broader labour market trends. After concluding its investigation, NISRA told us that it suspects an issue with the understanding of the question. The affected data were published and accompanied by an issue note. NISRA intends to revisit the purpose and wording of the question as part of future census planning. Given the insight gleaned from the Census Quality Survey on topic questions, we expect NISRA to use its 2021 results, in particular those questions with a lower than 85% agreement rate, to inform any future development of any survey questions, including the ‘Ever worked’ question.

Supporting the ongoing use of census data affected by the pandemic

2.15 Overall, Census 2021 data in NI are well trusted and regarded by users as being of high quality. The high response rates and NISRA’s extensive quality assurance process mean that users have confidence in using NI census estimates. The data from Census 2021 provide a snapshot of life in Northern Ireland at an unprecedented time. Data collected during this time may well be unusual or different from what might have been expected, and this was particularly the case for topics such as employment and economic activity. Interestingly, most users raised no concerns about the impact of the pandemic on either their use of or the quality of the statistics.

2.16 Supporting methodology documentation for the 2021 Census is available, with NISRA taking a proportionate approach to reflect the differing needs of users. For example, NISRA’s Census 2021 methodology overview summarises the census process, including the methods used, and provides links to more-detailed reports, such as the Census Quality Assurance Strategy (which includes details on the Census Address Register, methods used to process / validate responses, coverage estimation and adjustment and statistical disclosure methods). All the census topic bulletins and data tables also include links to enable users to navigate to additional methods documentation. The users that we spoke to were happy with the level of detail accompanying outputs. One user noted that the Flexible Table Builder could have benefited from further technical notes but understood the importance of not overwhelming non-expert users of the tool. Users were confident in using census data and content with NISRA’s methodology reports and supporting information.

2.17 NISRA included some explanation of the possible distortive effects in specific topic areas most likely impacted by the pandemic in its statistical bulletins. The labour market bulletin highlights the possibility that some people on furlough may have responded as being economically inactive instead of temporarily away from work. The bulletins for Migration and Place of work or study also highlighted to users that related statistics would be impacted by the pandemic owing to travel restrictions and the shift to home working for many.

2.18 There are limited explanations of how users should interpret trends over time and the extent to which decisions should or should not be made based on these data. NISRA told us that it is difficult to understand the full extent of the impact that the pandemic has had on the census estimates. To support the appropriate use of census data affected by the pandemic, NISRA should review its explanations and be more explicit in its guidance to help users understand what weight can be placed on affected statistics in decision making. In addition, signposting to other relevant topic statistics available in 2021 beyond those based on census could also provide users with a more holistic view of a particular topic. This work should be completed within three months of the publication of this report.

Meeting user needs and collaboration with other parts of the UK and the Republic of Ireland

Meeting user needs

2.19 Throughout the census process, NISRA has demonstrated that it understands the public value of the census outputs, and the overarching feedback from users was that the census outputs had met their needs. This was achieved in part through the two public consultations that NISRA conducted to gather input from users that fed into the design and delivery of the Census 2021 outputs. Users felt that NISRA recognised the range of potential uses for the statistics and that the census results had value in both describing society and aiding research and decision making. NISRA has a good understanding of the wide range of potential users from both within the statistical system and in wider society.

2.20 It is welcome that NISRA is planning to conduct a census benefits realisation exercise in 2025. Looking towards the future, one user suggested that NISRA could do a series of events and publications to showcase all the research and learning from the census. This would demonstrate that NISRA had also reflected on the public value of the outputs, not just the quality of the process.

2.21 NISRA demonstrated sensitivity and insight around how the data could be used to inform decision making and public debate, such as through publishing the guidance note on use of religion question outputs. Prior to Census 2021, NISRA carried out several community liaison events to increase awareness of the census among community groups and the voluntary sector. Users reiterated that they found it helpful to have an in-depth conversation with NISRA about their census experiences and concerns.

2.22 New questions on sexual orientation and renewable energy have been well received by users and, as new datasets, their results have been well communicated by NISRA. An outstanding gap raised by users was on education data in post-16 qualifications, where the lack of breakdowns in degree level and apprenticeship type limited the usefulness of the statistics.

Comparability with censuses in other parts of the UK and the Republic of Ireland

2.23 Another area where user needs were not currently fully met was in comparability with the Republic of Ireland and the rest of the UK. In general, comparability between Census 2021 and Census 2011 is high where the questions are broadly similar. Nonetheless, several users would like it to be easier to compare NISRA outputs with other contemporary datasets. Despite this, one user described the breakdowns presented in the downloadable tables of the outputs as being clear enough for them to conduct their own, albeit laborious, comparison work.

2.24 NISRA continues to actively collaborate with other statistics producers, particularly those in the UK and the Republic of Ireland, namely the Office for National Statistics (ONS), National Records of Scotland (NRS) and the Central Statistics Office (CSO). Building on those existing relationships, we would like to see the needs of all census data users met, particularly those interested in comparing NI census data with the rest of the UK and the Republic of Ireland.

2.25 While the census in the Republic of Ireland was held in April 2022, officials are still planning to produce a joint report on statistics from the 2021/22 censuses. This joint report is due to be published in 2025 and is widely anticipated by Northern Irish users of the Census 2021 results. The report will provide valuable high-level all-Ireland statistics on key areas covered by the censuses in Northern Ireland and the Republic of Ireland. NISRA’s strong collaboration with CSO is of significant public value and acknowledges the user interest in all-Ireland statistical outputs. NISRA has informed us that further work on comparability is also planned, in collaboration with other UK producers, which is being led by ONS. NISRA should continue to build on the work already planned with producers in the rest of the UK and the Republic of Ireland, as such cross-working would provide added public value to the availability and comparability of census statistics across jurisdictions.

2.26 UK statistics producers, led by ONS, are working towards providing UK census outputs. These efforts are formally supported by the Concordat on Statistics. The Inter-Administration Committee supports the implementation of the Concordat and resolves any inter-administration issues at a strategic level. Current plans and timings for UK census outputs are yet to be confirmed. Managing UK census data users’ expectations is important. To aid transparency and help manage expectations, NISRA should keep users updated on ONS-led plans to develop and publish UK-level estimates for key census categories. These updates should include information about anticipated timelines and what outputs are being developed in partnership with the Office for National Statistics (ONS) and National Records of Scotland (NRS).

NISRA’s user engagement

2.27 NISRA has taken an inclusive approach to supporting users to understand and make use of the census statistics. Its efforts here have included directly engaging with users and making a wide variety of resources and information publicly available on its website. Generally, users reported that the NISRA census engagement team had been effective in communicating with different sections of Northern Irish society and taking their requirements into consideration. Several users noted that the outputs had successfully engaged users and potential users, ensuring that data were publicly accessible. One user highlighted that more census-related information is available now than ever before.

2.28 NISRA used a variety of methods to engage with users about the Census 2021 outputs, taking a proportionate approach to increasing awareness of the data, reaching a wide audience in Northern Ireland. This approach has been sustained throughout the release timeline, thereby ensuring consistency and predictability for users.

2.29 The census team was described as helpful and professional. One user noted that the census team became less proactive in engaging with users once the outputs had been published. Another user suggested that NISRA could do more to signpost census-related events on its website to enable more public participation. NISRA told us that it was subject to resource constraints and needed to focus on other priorities at the time, such as providing research datasets. Further direct engagement with community groups following the release of the outputs could have reinforced the value of the statistics to the public and we suggest NISRA consider the benefit of any post census engagement in the future.

2.30 NISRA used a variety of methods to engage with users about Census 2021 outputs, including email alerts, social media posts, media briefings, press releases, webinars, custom presentations, visits to schools and universities and conference attendance. Government users and interested groups were supported through a series of well-received bespoke presentations about relevant outputs to Northern Irish government departments. NISRA has provided a comprehensive frequently asked questions page on its website. To aid user understanding, NISRA published information papers describing key elements of the Census 2021 methodology and data processing, with a dedicated area for reference materials.

2.31 To support sophisticated analysis of the data by the research community, NISRA is making more-detailed census information, such as microdata, available to accredited researchers. Examples include the Northern Ireland Longitudinal Study (NILS), the Northern Ireland Mortality Study (NIMS) and the Census 2021 Comprehensive Microdata Dataset (C21CM). Controlled access to specific microdata samples has been informed by consultation with users so that access to the most in-demand data can be prioritised. The release of microdata has conformed with requirements around confidentiality and disclosure control.

2.32 Supplementary statistical services are available through the Census Commissioned Output Commissioned outputs are made available to users on the census website. Users valued NISRA’s attention to these bespoke requests, including offering meetings to refine the user requirement. Subsequent statistical releases are consistently released in the same format and advertised on the same platforms. NISRA has published transparent correction notes in the form of Statistics Press Notices when necessary. At the time of publishing this report, NISRA is still planning to release additional Census 2021 products, including additional census microdata products for research.

Means of dissemination

2.33 The range of 2021 outputs and online tools produced by NISRA, in particular the Flexible Table Builder, has allowed users to maximise the value of census data for their own analyses and use. Several users with longstanding experience of using Northern Irish census outputs described the 2021 outputs as the most accessible census statistics published to date. By making thecensus outputs available in a range of formats, NISRA has demonstrated an understanding of the different types of users who may be interested in the statistics.

2.34 These tools are accompanied by statistical bulletins which have been published to suit users of varying The tools on the NISRA website are intuitive but are nonetheless accompanied by user guides. Ways of providing feedback are available to users to enhance NISRA’s insight into the user experience. This comprehensive approach to the user experience has been crucial in enabling public interaction with the statistics.

Website

2.35 NISRA published all of the census outputs and supporting materials on the 2021 Census section of its website. This enabled users to access all outputs through a single portal. Users reported that the website was a significant improvement on the Northern Ireland Neighbourhood Information Service used for the 2011 results, with accessibility and navigation greatly improved. Overall, the website is relatively simple to navigate, with links provided to the main statistics, Flexible Table Builder, commissioned tables and other output tools, as well as supporting materials, such as guidance, quality notes and revisions information.

2.36 The Main statistics tables provide summary statistics for the majority of census questions and topics, arranging published data by theme. The webpage also provides an overview of which statistics were released in each phase of NISRA’s release programme (Phases 1, 2 and 3). From here, tables can be relatively easily located and downloaded in a variety of formats. NISRA provides a bulk download option for users. Each table includes appropriate coversheet information to aid user understanding.

2.37 While most users found it easy to access the data they required, two users highlighted that the website could be more user-friendly and was at times difficult to use. Criticisms included that many pages on the website contained large bodies of text, which increased the cognitive burden on users, and that outputs were not always easy to navigate to from the landing page. Three users also noted that statistical bulletins were not always easy to find alongside relevant outputs, and that casual users may find them hard to locate. One user suggested that NISRA could improve the language accessibility of the outputs for minority groups, such as through using an imbedded automated translation tool.

2.38 NISRA has a Help viewing documents page which provides links to reader software. There is also the option to request alternative formats and guidance on PDFs. NISRA has an accessibility statement on their website and a specific accessibility statement for the Flexible Table Builder. While giving appropriate consideration to the needs of those with disabilities, NISRA has acknowledged that not all accessibility limitations can be overcome through these methods.

Online output tools

2.39 NISRA developed several online tools to help users access and interpret the statistics. These tools have increased accessibility and general interaction with the census outputs, helping to unlock public curiosity in the statistics. The Flexible Table Builder was a major innovation that enabled users to build their own census tables for the first time, using almost all categories and breakdowns from the census data. It maximised the utility of the census data, as millions of tables could be produced, thereby offering users much more flexibility. User feedback on the Flexible Table Builder was extremely positive, and its usability was vital to both technical and casual users of the outputs. Supporting explanatory information, such as training videos, was easy to find. This innovative tool has been a successful response to users’ request for a more flexible way of creating bespoke tables following feedback from Census 2011.

2.40 The Area Explorer, which allows users to explore geographies, was described by several users as intuitive. Its success led NISRA to develop the Northern Ireland Local Statistics Explorer. The continued publication of the Grid Square product, which allows for granular analysis of small area geographies, was also valued by users, though it is not as intuitive to use or visually palatable as the other output tools. An online version of the product, overlaid on mapping software, would likely be easier to navigate and use. The development of this product is planned under an ESRC funded project and is supported by NISRA.

Statistical bulletins

2.41 NISRA published 18 Census 2021 statistical bulletins to accompany the detailed characteristics tables. These provided comprehensive but accessible written summaries of the main findings of the census, offering users important commentary and context. They contained useful and appropriate data visualisations, such as maps and graphs, and objectively described the statistics, using straightforward language. Where necessary, the bulletins provided definitions and information about non-response and any uncertainty around the figures. Links were provided to a range of supporting documents, such as methodology, quality assurance and associated outputs.

2.42 The statistical bulletins are located on the release page for each phase of NISRA’s release programme and in a separate statistical bulletins This means that they are not always conveniently located alongside the main data tables and can be unnecessarily complicated to find. Given user feedback, NISRA should review the census section of its website to consider how navigability and the user interface can be further improved to enhance the user experience and increase the accessibility of the statistics.