The future of population and migration statistics: equality impact assessment

The Office for National Statistics (ONS) is committed to ensuring that the principles of equality and diversity are at the heart of all its decision making. To help achieve this goal the ONS has undertaken this equality impact assessment relating to its proposals for transforming how it produces population and migration statistics.

Completing this aligns with the requirements laid out by the public sector equality duty (the equality duty, PSED) as set out in section 149 of the Equality Act 2010. This requires public sector bodies to pay due regard to the need to:

• eliminate unlawful discrimination, harassment and victimisation, and other conduct prohibited by the Act

• advance equality of opportunity between people who share a protected characteristic and those who do not

• foster good relations between people who share a protected characteristic and those who do not

The nine protected characteristics are:

• age

• disability

• gender reassignment

• marriage or civil partnership

• pregnancy and maternity

• race

• religion or belief

• sex

• sexual orientation

In 2014, the UK government set out its ambition in the response to the National Statistician's recommendation (PDF, 100KB) that "censuses after 2021 will be conducted using other sources of data and providing more timely statistical information". In 2018, the ONS committed to responding to this ambition with a recommendation to government in the Census 2021 white paper 'Help Shape Our Future (PDF, 968KB).

Driven by this goal, the ONS has led a programme of research to improve its population statistics by further expanding the range of data sources it uses. It has shown that it can produce improved population estimates, and that it has developed methods for producing information about the population more often and more quickly. These methods will offer insight into our rapidly changing society as administrative data reach their full potential over the next decade.

The ONS believe that the proposed system would respond more effectively to society's changing needs and provide a more consistent level of detail and accuracy over time. If implemented, the system will give users high-quality population statistics each year. It will also offer new and additional insights into the changes and movement of our population across different seasons or times of day.

For many topics, it will provide much more local information not just once a decade but every year, exploring them in new detail and covering areas not recorded by the census, such as income. This is evidenced through the feasibility research into producing income by ethnic group statistics from administrative data at the Lower layer Super Output Area (LSOA) level.

Under these proposals, the ONS would provide richer, more frequent and timely statistics about the population, helping contextualise people's lives and enabling decision-makers across sectors to plan more efficiently and make better-informed decisions. The ONS would also be more sustainable and efficient by reusing, where possible, data that have already been collected through the provision of public services.

The current position

The census provides the backbone of population and migration statistics with a rich picture of our society at national and local levels every 10 years. Census outputs cover a vast range of topics including most protected characteristics.

In addition to univariate outputs (focusing on a single characteristic), explorable multivariate datasets (comparing multiple characteristics) are also made available. These allow users to combine and compare characteristics to gain an insight into how experiences differ between population groups. Accredited researchers also have access to the Office for National Statistics's (ONS's) Longitudinal Study allowing them to examine the interaction between characteristics and life outcomes.

Census data are complemented with limited survey and administrative data to reflect changes in society. Under this system, the ONS is less able to respond to rapidly changing user needs during intercensal periods than would be possible under the transformed system. Statistical outputs on protected characteristics between census years are often limited to regional level or are not produced at all.

Furthermore, levels of uncertainty in population estimates are known to increase over time in the intercensal period potentially leading to over- or under-coverage, particularly at local levels. This means the quality of information available to decision-makers is liable to decline across the decade.

The proposed system

The consultation document sets out in detail the proposed system. To summarise, the proposals comprise:

• improved annual population statistics that are released more quickly than they are currently

• estimates that allow the comparison of different attributes in the population at the local level provided much more often than currently possible, including for some protected characteristics

• broader analysis of outcomes over time via the Longitudinal Population Dataset

This would be achieved using administrative data and supplementary surveys, which would provide timely data on the population and protected characteristics. The ONS has shown that administrative data can be used to produce information about a range of characteristics. So far only a small part of the insight administrative data could provide has been realised; over time the ONS will work with partners across the public sector to expand the sources of data available and improve the range and relevance of data that is collected and used.

Considering a move to a system that is not reliant on a census would also allow the ONS the opportunity to revamp its approach to user engagement. This would mean it could respond more rapidly to requirements from different types of users for different topics. This would align with the desire to have rapid and responsive evidence on the matters of the day, as well as proactively focusing on wide-reaching emerging topics based on user needs. 

Trade-offs and risks

Each time a census is held, a questionnaire is provided to every household and communal establishment in the country, either electronically or on paper. For Census 2021 in England and Wales, the ONS received a 97% response rate. To account for non-response, the ONS uses a follow-up survey, other data sources and statistical methods to arrive at high-quality population estimates from census data.

The success of Census 2021 bucked an international trend, as response rates for all types of surveys have been falling for many years, and there have been examples of much lower response rates to censuses in other countries. It is possible the response rate to Census 2021 was so high at least in part because it took place during the coronavirus (COVID-19) pandemic when a high proportion of respondents were at home. While censuses tend to provide excellent coverage of the population, high response rates to censuses cannot be guaranteed in the future. If censuses receive low levels of response, the quality of the subsequent decade of statistics could be impacted.

The ONS's use of the dynamic population model (DPM) draws strength from the use of a broad range of administrative data sources, to avoid reliance on a single source. It's initial use in producing population estimates has already proven to be far more resilient when presented with interruptions in the supply of individual data sources than the methods used in the production of mid-year estimates (MYEs).

The ONS's proposed system would respond more effectively to society's changing needs and provide a more consistent level of detail and accuracy over time. If implemented, the system will give users high-quality population statistics each year. The ONS's research to date shows the potential for a net increase in statistical quality over the decade in its population estimates when using these new methods and sources. The ONS has also shown that administrative data can be used to produce information about the population.

The ONS continues to strive to identify the best sources of data across a broad range of population characteristics, including a number where data are not yet readily available from administrative data. For some topics, surveys would continue to play a role while the ONS works with partners to support administrative data to reach their full potential. Such topics include religion and sexual orientation, which are not routinely collected in administrative data sources currently used in ONS research or within ONS surveys.

Outputs from the transformed system would generally be more frequent and timely than can be produced currently. However, they may lack the level of detail given by decennial census outputs and the level of geographical granularity for univariate outputs. The data collected from the census are rich in detail as respondents are often able to write in responses where they feel they are not represented by tick-box response options.

For example, ethnic group data from administrative sources are usually restricted to the standard 5 or 19 tick-box response options, while census data provide a breakdown of 287 ethnic groups down to Middle layer Super Output Area (MSOA) [note 1] level based on write-in responses.

Next steps

This impact assessment is intended to be an interim document to complement the ONS's public consultation on the future of population and migration statistics.

The consultation exercise will provide the ONS's users with the opportunity to inform plans for the future of population statistics, including in relation to protected characteristics. This will also help inform the direction and prioritisation of ongoing research.

Responses to the consultation will be evaluated and considered by the National Statistician when making their recommendation on the future of population and migration statistics. An updated version of this impact assessment, considering consultation responses, will be published alongside the recommendation.

Notes for: Summary of the evidence considered in demonstrating due regard to the public sector equality duty:

1. Geographical Areas - OA -The lowest level of geographical area for census statistics, consisting of between 40 and 250 households (100 to 625 people). LSOA - Geographical area for ONS statistics usually comprising four or five Output Areas (OAs) and containing between 400 and 1,200 households (1,000 to 3,000 people). MSOA - Geographical area for ONS statistics, usually comprising four or five LSOA and containing between 2,000 and 6,000 households (5,000 to 15,000 people).

This section includes three subsections to consider impacts in different ways:

• Assessment against the public sector equality duty's (PSED) three aims

• Analysis of the impacts by individual characteristics

• Consideration of broader potential impacts

Assessment against the public sector equality duty's three aims

The three aims of the PSED require public authorities while carrying out their functions to have due regard to the need to:

• eliminate discrimination, harassment, victimisation and any other conduct that is prohibited by or under the Equality Act 2010

• advance equality of opportunity between persons who share a relevant protected characteristic and persons who do not share it

• foster good relations between persons who share relevant protected characteristics and persons who do not share it

By producing high-quality, timely population statistics, the Office for National Statistics (ONS) supports each of the equality duty's three aims and advances equality. Statistics about the population, in particular outputs broken down by protected characteristics, provide citizens and decision-makers with up-to-date and reliable information to help assess and address inequality, and provide a counter to misinformation and prejudice. Where the ONS can make improvements to the quality of these statistics, it means that the ONS and other bodies will be better able to meet the aims of the equality duty.

The consultation exercise will provide users of ONS data with the opportunity to inform the ONS's ambitions for statistics relating to protected characteristics.

Decision-makers including central government and local authorities use population and migration statistics to plan and allocate resources effectively to ensure people get the services and support they need, both within their communities and nationwide. Population and migration statistics describe the number and characteristics of people living in different parts of the country and how this changes over time. High-quality statistics enable robust and transparent decision-making and evaluation that helps prevent unlawful discrimination and advances equality of opportunity.

Understanding how things such as income and health can vary based on characteristics such as ethnic group and disability can help identify and minimise disadvantages, as well as show how the needs of particular groups may differ and enables those to be addressed.

Accessible and trusted information from the ONS on populations with protected characteristics and their lives improves public understanding of and between people from different groups. Accurate data foster good relations by providing society with the tools to tackle misinformation and prejudice.

One of the four core principles of the UK Statistics Authority's (the Authority) Strategy 2020 to 2025 is "inclusive". As the executive office of the Authority, the ONS is working to ensure that its future statistical provision will provide users with a more frequent and timely representation of protected characteristics to support the implementation of this part of the Strategy

The ONS's proposals to create a sustainable system for producing essential, up-to-date statistics about the population are set out in full in the consultation document. The ONS is proposing to use administrative data alongside survey data and a wider range of data sources to produce up-to-date statistics more often. This will mean that the ONS's statistics will be consistently accurate and detailed. The ONS proposes to deliver these outputs for protected characteristics sooner than statistics would next be available under the traditional 10-year cycle.

Under the ONS's proposals it would provide improved annual population statistics, which are released more often and more quickly than is currently possible. It would also provide more frequent multivariate estimates, which allow the comparison of different characteristics and attributes in the population at the local level; and a broader analysis of outcomes over time via a Longitudinal Population Dataset.

In some cases, there are decisions to be made in the short term about how to optimise the balance of frequency, timeliness, geographic granularity (that is, how small a geographical area statistics can be provided for), and detail of outputs (that is, the number of categories [note 2] statistics can be broken down into). Over time, these areas will be the focus of ongoing work to improve and develop the outputs to minimise this impact.

Analysis of the impacts by individual characteristics

This part of the equality impact assessment compares, for each of the protected characteristics, the statistics that are currently offered under the existing statistical system against what is expected to be available under the proposed new system, as well as plans for further development. Following the consultation exercise, this section will be updated to reflect information received from users.

Age and sex

Currently, the ONS releases estimates of the population by age and sex at the local authority (LA) level once per year, at the mid-year point, with a minimum 12-month lag from the reference date, and estimates with a 15-month lag at the Lower layer Super Output Area (LSOA) level of around 1,000 to 3,000 people. These mid-year estimates (MYEs) are constructed by rolling forward estimates from the census every year using information about births, deaths and migration. Error can build up in these estimates over time until a new census is carried out, when the previous years' MYEs are rebased (updated).

Under the ONS's proposals, it would produce "provisional" estimates of the population by age and sex in a timely way (six months after the reference date). It would then produce updated estimates relating to the same topics with a 12-month lag from the reference date.

These updated estimates would have a more stable level of accuracy over time than it can achieve through existing methods (currently, in the intervening years between censuses, the levels of uncertainty in population estimates increase). This approach would provide a balance of more timely provisional information, and more accurate information over the longer term. Initially both sets of estimates would be produced at the LA level, but over time the ONS would work to produce estimates at smaller areas covering geographies of interest, starting with the LSOA level, dependent on the quality of available input data.

The ONS's population statistics are based on "usual residence". As set out in the consultation document (section 3.1.1), it is exploring the use of administrative data to measure temporary resident populations and a diverse range of migration patterns. For example, seasonal migrants, or migrants who move into and out of areas, such as students, whether they are moving internationally or from one part of the country to another.

It is also looking at the information administrative data can provide about the "population present" over a given period, which can reflect much shorter stays in an area. The ONS expects that estimates based on these alternative definitions would be broken down by age and sex.

Race

Under the Equality Act 2010, race refers to "colour", "nationality" and "ethnic or national origins". Currently, the ONS produces outputs relating to:

• ethnic group

• national identity

• country of birth

to support monitoring of this protected characteristic.

Ethnic group

Based on census data, the ONS is able to produce ethnic group estimates every 10 years, with a lag of at least a year from the reference date. Univariate data for ethnic group are published to Output Area (OA) level; multivariate tables are also available and can be explored through Nomis and the Create a Custom Dataset tool, although depending on combinations of data.

Respondents are asked to identify their ethnic group from one of 19 tick-box options, grouped under five high-level classifications:

• Asian or Asian British

• Black, Black British, Caribbean or African

• Mixed or multiple ethnic groups

• White

• Other ethnic group (in Wales, "Asian Welsh" and "Black Welsh" were also included in the high-level descriptions)

The data available from the census are based on these 5 or 19 classifications; additional classifications are provided through the "Ethnic Group Detailed" release, which includes responses provided under the "Other" write-in options under each high-level grouping. These publications include 287 classifications to MSOA level.

ONS research shows that by making greater use of administrative data, there is potential for producing estimates of ethnic group down to LSOA level on an annual basis, with a 12-month lag from the reference date, across the 5 and 19 standardised classifications.

The ONS is working with partners to identify ways to improve data quality, including to adjust for the issue of missingness in administrative data; and to increase the number of breakdowns over time. Furthermore, the feedback received through the consultation will help the ONS to shape the future development priorities relating to ethnic group, national identity and country of birth.

National identity

The census for England and Wales asked a question on national identity in 2011 and 2021. Respondents were able to identity as English, Welsh, Scottish, Northern Irish or British via tick-boxes, or any other national identity (including others within the UK, such as Cornish) through a write-in box. This allows the ONS to produce estimates of national identity every 10 years.

As with ethnic group, univariate data are available to OA level for the tick-box response options and to MSOA level for the "detailed" response, which includes non-UK national identities. Multivariate tables are also available and can be explored through Nomis and the Create a Custom Dataset tool, although depending on combinations of data the available geographies may vary.

In future, subject to funding, the ONS will take forward research to build towards annual estimates of national identity with a 12-month lag down to the LA level.

Country of birth

The census for England and Wales asks a question on country of birth. Respondents are able to select England, Wales, Scotland, Northern Ireland or the Republic of Ireland via tick-boxes, or any other country through a write-in box.

Univariate data for national identity are available to OA level with responses grouped by region or continent (for example, European Union countries, or African countries) and to LA level for an extended list of 189 responses. As with other variables, multivariate tables are also available and data can be explored through the Create a Custom Dataset.

For this topic, surveys will continue to play a role while work with partners continues to identify the appropriate administrative data sources. Outputs for country of birth will be produced at a higher frequency than current census estimates but with a longer lag of around 15 months.

Disability

Disability is defined in the Equality Act 2010 as having a physical or mental health impairment that has a substantial and long-term adverse effect on the person's ability to carry out normal day-to-day activities. ONS data on disability are current based on that definition, reflected in the Government Statistical Service (GSS) Harmonised Standard on measuring disability for the 2010 Equality Act and the Census 2021 question on health and unpaid care.

Every 10 years, using census data, the ONS produces statistics with a two-year lag. Univariate data for disability are published to OA level; multivariate tables are also available and can be explored through Nomis and the Create a Custom Dataset tool although depending on combinations of data the available geographies may vary.

At present, the ONS produces disability outcomes estimates at the regional level once a year, with a six-month lag. This is produced by using data from various sources.

In response to the Inclusive Data Taskforce highlighting that there are currently significant evidence gaps surrounding disability data, work has been ongoing to rectify this. This can be evidenced in recent publications such as Protected characteristics by disability status in England and Wales and Ethnic group by general health, disability and unpaid care in England and Wales.

Under the proposals, administrative data and survey data would be used to provide provisional estimates of the population by disability annually down to the LSOA level. Through these means, the ONS would also seek to address the aforementioned evidence gaps.

Religion or belief

The ONS currently produces estimates on religion of the population on a decennial basis, with a lag of 12 months. This is based on data obtained via a voluntary question asked in the England and Wales census since 2001, in which respondents could tick one of six religions:

• Christian

• Buddhist

• Hindu

• Jewish

• Muslim

• Sikh

• "No religion" or "Other religion", for which a write-in option was available

As with ethnic group, univariate data for religion are available to OA level for the tick-box response options and to MSOA level for the "detailed" datasets, which includes 50 other responses; multivariate tables are also available, and data can be explored through the Create a Custom Dataset tool.

The proposed system will primarily use survey data to produce these estimates on an annual basis down to LA level. Because of reliance on survey data it will be difficult to deliver estimates that go below the LA level. However, further research is required to identify how much of a time lag there will be between the collection and the production of the estimates. Feedback received through the consultation will help the ONS to shape the future development priorities relating to religion and belief.

Sexual orientation

Sexual orientation estimates for the population aged 16 years and over are currently produced on an annual basis at the national and regional level. These are produced using the Annual Population Survey. The estimates are currently produced with a lag of around 10 months.

For the first time in 2021 a voluntary question on sexual orientation was asked in the census, providing the most detailed estimates on this topic to date, available down to MSOA level for univariate sexual orientation data and lower tier local authority level for many multivariate sexual orientation datasets.

Under the proposed changes the ONS will use a different question within the Transformed Labour Force Survey to produce wider estimates on sexual orientation on either an annual basis at a more granular level or on a quarterly basis at more aggregated levels. This will be made possible because of a larger sample being surveyed. The current anticipated lag for the production of these estimates is over 12 months.

Gender identity

While the protected characteristic in the Equality Act is gender reassignment, the gender identity question in Census 2021 in England and Wales provided an estimate of the size of the population who might be liable to experience discrimination or differing life outcomes in direct relation to that characteristic. This was a voluntary question, asking those aged 16 years and over whether the gender they identify with is the same as their sex registered at birth. Those answering "no" were asked to write in their gender identity.

The ONS has now published its first research report Update on research into gender identity data in Census 2021, which examines whether the information that people provided on the census was coded and processed accurately. The report concludes that 94% of respondents to the census answered the question and there is no evidence to suggest that the published results were not coded or presented accurately. Further ongoing research is taking place to compare census gender identity data with other available sources and to examine what data patterns and other information collected on the census can suggest about how people answered this question.

Census 2021 provided the first official estimates on this topic, available down to MSOA level for univariate gender identity data and lower tier local authority level for many multivariate  gender identity datasets.

The ONS does not have access to administrative data relating to gender identity at present. The proposed new system will use data on gender identity from surveys and potentially administrative data to produce estimates at a biennial frequency. This will be possible down to the LA level with the expectancy of a lag longer than 12 months from the reference date.

Marriage and civil partnership

The census provides estimates every 10 years based on self-reported marital or civil partnership status. These data from Census 2021 can be explored alongside other variables through the Create a Custom Dataset tool.

At present, estimates are produced on marital and partnership status on an annual basis at the national level. Data from this are compiled from census data and survey data as well as data from the General Register Office. The figures are produced on an annual basis and there is currently a time lag of six to nine months between the data reference date and producing the publication.

Marriage estimates are produced annually at the Registration District level and LA level with a time lag of two years between the data reference date and producing estimates. Data are compiled from the General Register Office and from the International Passenger Survey. The latter is used to estimate the number of UK residents marrying abroad and overseas residents marrying in the UK.

Civil partnership formation estimates are also produced annually at LA level but with a time lag of approximately two years. These estimates use data from the General Register Office.

Divorce and civil partnership dissolution estimates are produced annually for England and Wales as a whole using data from HM Courts and Tribunals Service.

Further, the ONS produces population estimates by marital status and living arrangements annually at national level with a time lag of 13 months. These estimates are produced by applying data from the Labour Force Survey to mid-year population estimates.

For the proposed new system, the ONS intends to explore how it can exploit the rich information already available from the General Register Office alongside administrative data sources provided by other government departments to improve information about marriage and civil partnerships.

Pregnancy and maternity

Data on this topic are not collected through the census. Instead, the ONS produces statistics on conception and births using administrative data sources. These are birth registrations from the General Register Office, abortion notifications from the Department of Health and Social Care, and NHS records.

Conception estimates are produced annually to LA level. These have a time lag of 14 months from the end of the data year. These estimates are broken down by age and ethnicity. Additionally, conception estimates for women aged under 18 years are produced quarterly to LA level also with a time lag of 14 months.

Birth estimates are produced annually with a six-month lag at LA Level. The estimates are broken down by country of birth, age and ethnicity of the child, marital status of the parents, socio-economic classification of the parents and the number of births to same-sex female couples.

The proposed changes would have a more stable level of accuracy over time than can be achieved through existing methods. This will provide a higher level of accuracy and timeliness because of using up-to-date data.

Multivariate analysis and cross-tabulation

The census provides a very detailed set of records from which many characteristics can be compared, once per decade. In the intervening period, the ONS must rely on data from the previous census, which can be up to 12 years out of date (for example, the latest figures in January 2023 were still using Census 2011 data until new data were released in February or March 2023), and surveys that do not allow the level of granularity in terms of geography and population groups.

The ONS ambition is that by 2033, it will be possible to conduct multivariate analysis across the same variables that census collects, once per year, with a much shorter lag. However, for certain combinations of variables, this may be at less granular levels than the census offers. This would start at LA level, building towards MSOA or LSOA level when possible.

The ONS intends to understand the impact of change here through its public consultation in summer 2023.

Consideration of broader potential impacts

This part of the equality impact assessment considers the potential broader equality impacts of transformation and describes how the ONS is actively managing these risks.

Understanding the representativeness and inclusivity of data sources

Understanding the quality and coverage of data sources is vital to producing inclusive and representative statistics. No survey or administrative data source provides a 100% coverage of the population, so for all population statistics the ONS uses estimation methods. These estimation methods would share similarities between current and future systems. The ONS recognises that representativeness and inclusivity varies in some new sources, including for some protected characteristics, and it has undertaken a range of activity to account for this.

Understanding coverage and linkage quality of administrative data

In order to better understand the level of coverage and linkage quality between data sources, the ONS has researched this by securely linking data from the census, the Census Coverage Survey (CCS) and the ONS's administrative data about the population. This has helped the ONS to optimise the inclusion rules for the Statistical Population Dataset, which will form the basis of its admin-based population estimates under the proposals.

New methods to understand inclusivity and representativeness

The ONS is developing a range of new methods to measure quality, inclusivity and representativeness in administrative data for statistical and research purposes.

The ONS has carried out innovative qualitative research to understand administrative data inclusivity and representativeness at the start of the data journeywhen members of the public are registering their information for access to services.

The ONS is also collecting qualitative information from administrative data suppliers to understand more about different aspects of data quality, including inclusivity and representativeness in administrative data.

Additionally, the ONS is developing the use of multiple imputation latent class modelling (MILC) for administrative data aiming to develop these methods further.

The findings from the ONS's programme of research provide evidence of inclusivity and representativeness that it will use as indicators of quality in administrative data. It will also feed into the ONS's existing and developing methods; and feed into its series of toolkits and guidance on administrative data quality that the ONS developed for cross-government use. This work will result in a series of quantitative and qualitative quality indicators that can be used to assess quality; and where possible, inclusivity and representativeness in administrative data for research and statistical use.

Managing conflicting information in administrative data sources

In research to produce admin-based estimates of ethnicity, it was identified that information about people in different administrative sources was sometimes inconsistent. With the advice of the National Statistician's Data Ethics Advisory Committee (NSDEC), the ONS has researched approaches for dealing with multiple recorded ethnicities and conducted focus groups to test public attitudes to different options. Based upon this work the ONS has developed selection rules to help manage inconsistency in the data and maximise the quality of admin-based ethnicity outputs. Research to improve these statistics further is ongoing.

Balancing granularity with statistical disclosure control

The ONS applies statistical disclosure control methods to prevent the identification of individuals within aggregate statistics. Current agreements around statistical disclosure control with suppliers of administrative data prevent the ONS from producing admin-based statistics about some characteristics to more granular geographical levels. Based on responses to the ONS's consultation in summer 2023, the ONS will review whether these arrangements should be renegotiated.

Levels of trust and how this affects representation in data sources

Privacy, security and ethics are fundamental to building and maintaining levels of public trust in the ONS's processing of personal information. Consequently, the ONS has sought advice regarding its plans from the Government Legal Department, the Information Commissioner's Office and the National Statistician's Data Ethics Advisory Committee (NSDEC).

On the basis of this engagement, the ONS has conducted qualitative research to ensure its methods for producing estimates based on administrative data are acceptable to the communities these estimates describe. In addition, a Data Ethics Self-Assessment Tool has been used before the starting of all research and this will continue for further research.

To increase the levels of trust in the use of administrative data and to ensure that a high level of trust in the ONS is maintained, the ONS has conducted the following activity.

The ONS has conducted roundtable discussions with a series of stakeholder groups, including internal stakeholders, local authorities, other government departments, charities, privacy groups and more to communicate its transformation plans to a broad audience and receive feedback. The ONS's public consultation in the summer of 2023 will provide the opportunity for further feedback on the ONS's proposals. Information is also available on the ONS website to explain its transformation plans to the general public, and video content with this information has also been produced for sharing via social media. This will include a video on the methodology of the DPM to make this understandable to the public and non-expert users.

Data protection has been at the forefront of the ONS's research and development, with data protection impact assessments produced for all research.

The engagement approach will ensure that bespoke engagement and communications are undertaken with specific groups ensuring a wide and considered reach is maintained. There are plans to engage with the main groups who have lower trust in the ONS during 2023. This includes working directly with those aged 18 to 24 years to test how this specific group responds to issues around data use. The ONS is also working with others across government and through the Public Engagement in Data Research Initiative (PEDRI) to promote public understanding of data and are launching a PEDRI pilot campaign in 2023 aimed at new parents, another group with lower trust.

Digital literacy, online-first collection and dissemination

The proposed new system will rely on an online-first approach to survey and data collection. Where surveys are still required there will be a risk that this may exclude some respondents. To mitigate this risk there has been consideration to two main areas: data collection and the publication of ONS data.

Data collection

To mitigate the risk of excluding people from responding with an online first approach the following approaches have been adopted. Development in this area will continue to ensure that people are not excluded:

• optional telephone surveys - for those without internet access who are asked to respond to a survey, it is possible to conduct the survey via telephone instead

• targeted populations - to ensure that the surveys reflect the population and provide a representative picture; target populations are being identified and included in the survey samples

• respondent-centric survey design - work has been undertaken and will continue to evolve ensuring that the surveys are designed with the respondent in mind; this means that the surveys are made to be accessible and inclusive

• translation services - interpreters are now offered with telephone interviews, providing assistance with translation for up to 250 different languages; this ensures that a language barrier does not exclude people from participating in surveys

The ONS's ongoing research into quality is also exploring ways to increase inclusivity in surveys.

Publication of ONS data

The ONS will ensure that outputs and products that are produced continue to meet its high standards for accessibility and inclusivity.

 For example, the ONS website allows users to:

• change colours, contrast levels and fonts

• zoom in up to 300%

• navigate most of the website using just a keyboard

• navigate the website using speech recognition software

• listen to most of the website using a screen reader

Furthermore, users are able to request information, documents and outputs from the ONS website in a different format if what is already provided is still not accessible.

Feedback is actively encouraged from all users to help ensure that the publications from the ONS remain accessible and evolve with technology and needs as appropriate.

Notes for: Identification, analysis and decisions of the impacts:

Data for some characteristics are typically collected by selecting the relevant tick-box answer, but the census also offers the option to write a response for some characteristics. The number of categories these data can be broken down to depends on the number of possible responses in these write-in boxes alongside tick-boxes; whereas for admin data this number is frequently limited to the number of tick-boxes alone.

Following the consultation, the National Statistician will make a recommendation on the future production of high-quality population and migration statistics that serve the public good, including a recommendation on the future of the census in England and Wales.

A final version of the equality impact assessment will be published alongside the recommendation. It will reflect comments from users made during the consultation and engagement exercises along with any updates and progress that has been made in the development of the proposed new system.