Dear Mrs Miller,

While providing evidence at the Women and Equalities Committee on 6 February on preparations for the 2021 Census, Frankie Kay and I promised to provide further information on a number of areas. Most relate to our recommendations on questions for inclusion, and as I discussed at the Committee, our role is to ensure that we produce high quality estimates of the population. To assess topics for inclusion in the Census we set out criteria which we then use to assess existing and new questions. These criteria include user need and public acceptability.

This letter follows up on: intersex data and gender identity; maternity data; data for the Sikh community; Census outreach; the legislative process and the future of the Census.

Intersex data and the sex and gender identity questions

The Committee were interested in more detail on the testing and process carried out for collecting information on intersex in the 2021 Census.

The sex question in the Census is an essential variable with high levels of user need that feeds into population projections, decision-making, planning and resource allocation across central and local government. This data reports toward a protected characteristic in the Equalities Act 2010. However, we know that some individuals in the intersex population might still wish to identify as being intersex through the Census.

From the beginning of our consultation work in 2016 we actively sought out and invited engagement from intersex organisations, including Intersex UK and the UK Intersex Association (UKIA), and we have kept these two groups informed of our question development work. Since 2018 we have also had discussions with: DSD Families, Turner Syndrome Support Society, Educate and Celebrate, Living with CAH and Agender.

We tested a change to the sex question, notably the addition of a third category for ‘other’. We ensured those who are intersex were included as respondents in the testing and development of this question. Our testing identified a number of different views. Intersex organisations advised that responding to the sex question via a third category was potentially difficult as those who are intersex have often experienced stigmatisation and might not be comfortable with disclosing their sex, while some individuals noted that the current question was a barrier due to the binary categories. Testing found that, overall, the addition of a third option meant individuals were unclear how to answer and the quality of male and female data was reduced. Our stakeholder engagement activity also uncovered some very strong views around the acceptability of adding a third response option, and that the sex question should remain as a binary option.

While we recognise that intersex is not the same as gender identity, there is a free-text box on the gender identity question where an individual can write in ‘intersex’ if they would like, meaning that some data on intersex may be collected through the Census. There will be guidance to ensure those who wish to inform us they are intersex are able to do so and know how to. Our engagement with the intersex community has not shown any objection to our proposed approach.

For those who identify as non-binary, we have recommended adding a caveat to the sex question that a gender identity question will follow. This makes the sex question more acceptable to non-binary and transgender people. The gender identity question will include a write in response option where non-binary respondents can identify as such. We are currently developing our plans for the processing of sex and gender identity data across both paper and electronic versions of the questionnaire. These will build on previous methods, but also consider that the Census will be conducted predominantly online in 2021.

The Committee might also be interested to note that National Records of Scotland (NRS) published ‘Plans for Scotland’s Census 2021’ in September 2018, where they set out their current proposals for Scotland’s Census 2021. They announced that they were considering a change to the response options for the sex question, and that testing was continuing on a non-binary question which would allow people to record their sex as female, male or other.

The Scottish Parliament’s Culture, Tourism, Europe and External Affairs Committee published a report on the Census (Amendment) (Scotland) Bill on 7 February 2019. It stated: ‘the Bill has raised significant issues with regard to how the Census in 2021 will deal with questions regarding sex, gender identity and sexual orientation and the process of consultation which has underpinned policy development on these issues’.

The Culture, Tourism, Europe and External Affairs Committee noted that the approach in England and Wales in 2021 will be to continue with a binary sex question and that the Office for National Statistics (ONS) Equality Impact Assessment underlined the importance of the protected characteristic of ‘sex’ to ensure robust data-gathering. This view was also expressed by data users in evidence to this committee.

The Culture, Tourism, Europe and External Affairs Committee recommended that the mandatory sex question should remain binary to maximise response rates and maintain longitudinal consistency with previous Censuses.

Pregnancy and maternity data

The Committee asked for further detail on why pregnancy and maternity data would not be collected in the Census. The Census is a snapshot of the population every ten years, and accordingly there was little user need for this information from the Census specifically which came out of our consultation process. Even without specific questions, the Census is still a valuable source given those who have recently given birth should be identifiable. It does provide other insights as well: by combining the responses from several questions, we can see if the mother of a child under 26 weeks are in the same household. We can also identify from the employment status question whether an individual is on maternity or paternity leave.

As we discussed at the Committee, this is an area where administrative data is most useful. For maternities, we have the civil registration record of every birth and stillbirth that takes place (and is registered) in England and Wales. We also have birth notifications from the health service, which give gestation length, baby’s weight, ethnicity, and we link these to the birth registration to produce our statistics on births.

These data on essentially every birth and mother gives us a good basis for linking to other administrative or survey data to generate new insights. This could be health data about the mother, her pregnancy, delivery, and post-natal care. It could also be non-health data on things like income and employment before and after the pregnancy and birth.

We are in the process or exploring and acquiring new health and non-health administrative sources. Moreover, our health analysis team’s wider strategy is to explore the social, economic and environmental impacts on health, but also how health impacts these areas. There will be challenges with the administrative sources we acquire, but their acquisition in tandem with our existing data and analytical strategy, should allow us to explore research questions such as those around maternity and equality, without any need for Census data on pregnancy or maternity.

Survey of Gurdwaras

At the Committee we briefly discussed the survey of Gurdwaras. A possible survey of Gurdwaras was first mentioned at a meeting on 17 January 2018. The meeting was convened by ONS with leaders of the Sikh community to discuss how best to meet the data needs of the Sikh community in the 2021 Census. There were differing views between the Sikh leaders on whether a new tick-box for Sikh ethnicity was required in addition to the tick-box for Sikh religion. It was agreed that the meeting would reconvene in a months’ time and that in the intervening period, ONS would seek legal advice from the equalities perspective. It was also agreed that if, at the next meeting, there was still no obvious way through, a survey of Gurdwaras could potentially be a way forward. A covering letter and a survey was issued to Gurdwaras by the APPG for British Sikhs dated 31 January 2018. The first I was aware of the survey was when I received a letter from the APPG for British Sikhs with a copy of the survey attached on 12 February 2018. I neither saw nor commented on the survey prior to this date. On 19 February 2018, I replied to the APPG for British Sikhs to say it was my understanding that they would share material to support a potential survey, rather than sending a survey out directly. I explained there are important considerations to ensure surveys are unbiased, and that ONS would utilise our independent statistical assurance panel to assess whether the survey will provide unbiased results. It is standard practice in ONS to seek expert statistical methodological advice for survey work. The independent advice which ONS received on 27 February 2018 said that:

• the survey was likely to be a survey of Gurdwara leadership, rather than the wider community;
• the covering letter put a strong case for the inclusion of the ethnic question, and then asked for the survey to be returned. This is not conducive to the collection of independent statistical evidence; and
• if there is a positive response to the survey, that does indicate support of a “political” or “advocacy” nature from the leadership of Sikhs.

Despite these reservations, on 10 April 2018 and again on 31 May 2018, ONS requested the results of the survey so that it could take it into account in finalising its recommendations. An article was published in the Times on 23 July 2018 announcing the results of the survey. ONS received the results of the survey from the APPG for British Sikhs on 13 August 2018, although the letter was dated 27 July. The results showed that:

• Of 112 Gurdwaras who responded, out of 250 Gurdwaras who were sent a letter, all supported a tick-box;
• The response rate from the Gurdwaras was less than 45 per cent.
While ONS did not endorse the survey or provide comment in advance on methodology, we informed the APPG for British Sikhs that it would still form part of the information we used to make our final recommendations, as is the case for all information we receive.

Census outreach

The main aim of the Census community engagement programme is to improve responses, particularly among the key population groups.
The programme develops strategic partnerships with organisations in the community, the voluntary sector and local authorities. It is focused on building alliances based on a shared interest to ensure communities are represented in the Census. We also recognise local authorities’ knowledge and understanding of their areas and resident communities which will, as in 2011, be important to the success of the 2021 Census.

The 2011 Census showed that through the development of these partnerships, ONS could access established and effective networks and communication channels to disseminate messages and share content.

The budget for this outreach work funds a small in-house team of six permanent ONS staff, who identify and establish these strategic partnerships with local authorities and national community organisations. They deliver face to face engagement with organisations and individuals and have established partnership forums. For example, they recently attended a Nepali community forum to raise awareness of the Census and met Age UK to explore how we can best support more vulnerable people in society to complete their questionnaire. The majority of content and promotional materials they provide to partners is produced in-house.

From summer 2020, a network of 150 temporary local Census Engagement Managers will be in place to build on these links in local areas. In addition to their work supporting operations, much of their focus will be on engagement with key population groups that are less likely to respond without support.

Our experience from the 2011 Census was that many local authorities, national and local communities, networks and organisations were motivated to help and conducted many engagement activities themselves, which contributed to the success of the Census. The outreach work aligns closely with our plans on engagement through the Census field staff and our communications campaign.

Legislative process

I promised the Committee I would provide a fuller explanation of the parliamentary process for the Census. We are working closely with the Cabinet Office on the legislation needed to deliver the 2021 Census as set out in the White Paper, with questions on sexual orientation and gender identity. Options include the potential for primary legislation, as was discussed in the evidence session, and which is the approach which Scottish Parliament are taking, however I should be clear to the Committee that the approach to be taken for England and Wales has not yet been finalised.
The Committee asked specifically about the scope for Parliament to amend Census questions. All Census legislation remains subject to drafting, however the content and conduct of the Census are each subject to approval by Parliament through a Census Order and subsequent Regulations.

The Census Order is an unusual statutory instrument, as the Census Act 1920 gives Parliament scope to amend certain aspects of the Order, but not others. The Census Order sets the date for the Census, as well as what information will be required from respondents, which must all relate to a list of topics that are listed in the Schedule to the Census Act 1920. The list in the Schedule is as follows:

1. Names, sex, age
2. Occupation, profession, trade or employment
3. Nationality, birthplace, race, language
4. Place of abode and character of dwelling
5. Condition as to marriage (or civil partnership), relation to head of family, issue born in marriage
5a. Religion
6. Any other matters with respect to which it is desirable to obtain statistical information with a view to ascertaining the social or civil condition of the population

The Census Act 1920 prescribes that only matters falling under paragraph 6 of the Schedule (‘Any other matters with respect to which it is desirable to obtain statistical information with a view to ascertaining the social or civil condition of the population’) are subject to the affirmative procedure and may be amended. Matters falling under paragraphs 1 to 5A of the Schedule cannot be amended.
The Census Regulations follow on from the Order and expand on it with further detail, but are only subject to the negative procedure and are not amendable.

The Order for the 2021 Census remains subject to drafting, however further to the Act itself, the 2009 Census (England and Wales) Order as laid before Parliament in draft provides a general guide to where there is precedent for questions to be amendable and where not. At this stage, it is too early to say how new questions will be treated in the Order.

The Census beyond 2021

Decision-makers increasingly need more granular and timely national and local data on the size, characteristics and changing dynamics of their population to build better services, transport links, schools, hospitals, housing etc. We spoke at the Committee about our work to transform population statistics by putting administrative data at the heart of the system in 2020. This and the 2021 Census are key to laying the foundation for us fully transforming to a statistical system based firstly on administrative data supplemented with surveys.

We will consult with users on how administrative data can come to the fore of the statistical system and ensure we are ready to make recommendations to Government in 2023 on the future of the Census. A research engagement report on the transformation of population and migration statistics has been published in January 2019.

An Administrative Data Census would produce population and migration statistics using data which other government departments already collect, for example, NHS patient registration. The latest assessment on progress towards this, published in July 2018, reported that ONS is making good progress but that further work needed to be done. Since then, we have developed a methodology for estimating household size and household composition from admin data and published a wider range of research into producing population characteristics, including new outputs on new mothers’ income, internal migration, household composition, ethnicity, small-area income distributions, commuting flows and labour market status.

We intend to produce new research outputs on:
• Educational qualifications (by end-2019);
• Ethnicity for small areas (using integrated Census and admin data);
• Overcrowding estimates using Valuation Office Agency (VOA) data and research into use of property type, with edit and imputation methods for number of rooms in preparation for the 2021 Census;
• and our progress on transformation of the population and migration statistics system.

We also hope to produce research outputs on income distributions (including Self-Assessment data) and labour market status by the end of 2019 if the data required becomes available. We will continue to keep the Committee informed about this work.

I hope the Committee finds this note helpful. We will of course keep you updated on our plans for the 2021 Census as they develop, and I would be very happy to have a meeting with yourself and any other members of the Committee to discuss this letter in more detail. Please do not hesitate to contact me if you have any further questions.

Yours sincerely,

Iain Bell
Deputy National Statistician and Director General, Population and Public Policy Office for National Statistics