Demography question development for Census 2021

In December 2018, the government presented to Parliament a White Paper Help Shape our Future: The 2021 Census of Population and Housing in England and Wales (PDF, 967KB). This outlined the Office for National Statistics’ (ONS’s) proposal to collect information on date of birth, sex, marital and civil partnership status, and household relationships to meet the requirement for information on the population structure to understand for example future hospital, school, and housing needs.

We will use the 2011 Census question topics on demography for Census 2021 and collect these data in a way that is consistent with previous censuses. The questions used to collect data on demography are:

• date of birth
• sex
• marital and civil partnership status
• household relationships

The question on sex is not discussed in this report; information on the development of this question is in the Sex and gender identity question development for Census 2021.

Census 2021 will be an online-first census, with a target of 75% online returns. We have reviewed and tested each question online and have recommended additional design and functionality amendments specific to the electronic questionnaire. We have also recommended additional changes to the question designs following our stakeholder engagement, evaluation and testing programmes. These changes will be detailed in this report.

The main recommended changes to question designs are:

• the marriage and civil partnership question, will be a new two-stage question design in Census 2021, whereby the first stage collects the legal status and the second stage asks if the respondent is, or was, in a marriage or civil partnership with someone of the opposite or same sex

• we have updated the marriage and civil partnership question to reflect that civil partnerships will be available to opposite-sex couples by Census 2021

• in the household relationship question, we have updated the response option “Same-sex civil partner” to “Legally registered civil partner” and added instructions to include half-brother or sister to the “brother or sister” response option

• we have added extra guidance on the paper questionnaire to aid respondents answering the relationship matrix grid on paper

• the household relationship question collects relationships between household members between every five people on paper and for the electronic questionnaire, which was previously every six people

• the paper questionnaire captures the relationship between household members through a matrix, whereas on the electronic questionnaire this information is collected through page by page questions and response options, to meet government digital service’s service standards

Any additional changes to question designs, including online functionality and accessibility, will be detailed in the sections Research that led to the 2018 White Paper recommendations and Research that led to the recommended Census 2021 question designs.

Since the publication of the White Paper, the Office for National Statistics (ONS) has conducted and concluded the final phase of testing on the demography questions.

This report provides links to previously published research and the findings of additional testing that led to the final recommended questions for Census 2021 in England and Wales. The questions and response options for Census 2021 have now been finalised through the census secondary legislation: the Census (England and Wales) Order 2020, and Census Regulations for England and for Wales.

The evidence base for the recommendations made in the White Paper and the finalisation of the date of birth questions for Census 2021 is discussed in the section Research that led to the 2018 White Paper recommendations. Further research into the relationship matrix and marital status questions are discussed in the section Research that led to the recommended Census 2021 question designs.

In June 2015, the Office for National Statistics (ONS) held a formal, 12-week consultation process asking census users for their views on the topics that were required in the questionnaire in England and Wales. The aim of the consultation was to promote discussion and encourage the development of strong cases for topics to be included in Census 2021.

In May 2016, we published our response to the 2021 Census topic consultation. This set out our updated view on the topics to be included in Census 2021, including:

• a summary of proposals for new topics
• next steps
• an overview of our plans

A detailed summary of the consultation responses relating to the topic of demography can be found in the Demography topic report (PDF, 1.6MB). In this report, we made clear commitments to the public. These were to ensure the marital and civil partnership status questions were kept up-to-date and in line with current legislation, such as the Marriage (Same Sex Couples) Act, 2013 and the Civil Partnerships, Marriages and Deaths (Registration Etc.) Act 2019. Also, to improve the respondent experience, reduce the respondent burden and investigate ways to improve data quality for the marital and civil partnership status question.

We have provided an update on how we met these commitments in Annex 1.

Overall, respondents to the consultation reported that the census is their main source of detailed information on each of these topic areas, which informs vital population estimates and projections and therefore underpins decision making, planning and resource allocation across central and local government. The information collected from these topics also effectively informs targeting of services and resource allocation for specific groups within the population. Additionally, age, sex, and marital and civil partnership status are protected characteristics under the Equality Act 2010.

Following this, we began a comprehensive programme of research and development. We provide a full list of the tests used in the development of the topic of demography in Annex 2. Further details are provided in the summary of testing for Census 2021.

The tests utilised a range of qualitative and quantitative research methods. A short description of the different research methods and sampling techniques is given in the question and questionnaire development overview for Census 2021. Testing included respondents from a wide range of backgrounds. This included differing marital and civil partnership statuses and ages.

In December 2017, we published a further census topic research update, which highlighted progress of the work on the legal marital or civil partnership question, and on the relationship matrix question.

In December 2018, the government presented to Parliament a White Paper Help Shape our Future: The 2021 Census of Population and Housing in England and Wales (PDF, 967KB). This outlined our proposals for the topics of date of birth, marital and civil partnership status, and household relationships. These were to:

• collect information on name, sex and date of birth
• collect information on family relationships for households with two or more people
• conduct testing to review the marital status question and response options to ensure all the necessary data are collected and seek to ensure that all legally recognised statuses are covered
• consider next steps for the marital and civil partnership status question response options in the light of the government’s response to the Supreme Court judgement concerning the availability of civil partnerships to opposite-sex couples

Alongside the White Paper we published a further Census topic research update. This provided additional details of the research that supported the demography recommendations announced in the White Paper.

The question recommendations for Census 2021 are now finalised. We have evaluated the questions on demography for potential impact on data quality, public acceptability, respondent burden, financial concerns and questionnaire mode. We present details of this evaluation in Annex 3.

Annex 4 provides details of the definitions and terms used in this report.

The starting point for reviewing the questions was the 2011 Census question designs. You can see these questions on the 2011 Census Household Questionnaire (PDF, 1.8 MB).

As set out in the White Paper Help Shape our Future: The 2021 Census of Population and Housing in England and Wales (PDF, 967KB), with further details provided in the December 2017 and December 2018 census topic research updates, we reaffirmed our commitment to asking questions on date of birth, marital and civil partnership status, and household relationships.

This section provides the details on how the questions on date of birth evolved from the beginning of testing through to the final recommended questions for Census 2021.

References to tests are provided in the form (Year: Test number). For example, the fifth test conducted in 2017 is referenced as (2017:5).

Date of birth

A question on age has been included since at least the 1841 Census (the first to ask detailed information about individuals) and date of birth since the 1939 Identity Card Registers for England and Wales.

There are no changes to the 2011 Census date of birth question stem on the paper questionnaire for Census 2021.

2011 Census date of birth question

What is your date of birth?
Day    Month    Year
[ ]      [ ]      [ ]

Electronic questionnaire

The electronic questionnaire date of birth question will use validation to ensure a valid date has been entered. The earliest date that can be entered is 21/3/1906 (115 years old), the latest date that can be entered is census day 21/3/2021. Additionally, there is an example text above the data entry fields that advises respondents on the correct format to use to supply their date of birth:

“for example, 31 3 1980”

Confirmation of age question question

Date of birth is crucial for the collection of demographic information critical across a wide range of census uses. Age is derived from date of birth and derives the age at a person’s last birthday (mid-year and census day). The variable informs population estimates and projections and therefore underpins decision making, planning and resource allocation across central and local government. Age is vital for multivariate analyses for the vast majority of users and is a protected characteristic under the Equality Act 2010.

In the 2011 Census, the agreement rate for date of birth question was 98.4% (single year) and 99.3% for five-year age bands, suggesting there was some inaccuracy in date of birth information. It is known in survey design that some people use “today’s” date for their date of birth in error, or mis-calculate the year of birth, particularly when answering on behalf of someone else.

Question design for online

In addition to the testing described earlier, the demography questions have undergone significant user experience (UX) testing (2017:2, 2018:2, 2019:1, 2020:2). UX testing focuses on understanding user behaviours as people interact with online services. Through observation techniques, task analysis and other feedback methodologies, it aims to develop a deep knowledge of these interactions and what it means for the design of a service.

UX research has taken place on a rolling, ongoing basis since 2017 and the programme will continue through 2020. All participants are purposively selected to include a wide range of ages and digital abilities across the country.

UX testing included various iterations of the demography questions described elsewhere in this report. Feedback from this research informed decisions made on the design of these questions. For more information on UX testing, see the Question and questionnaire development overview for Census 2021.

We tested adding a confirmation of age question on the electronic questionnaire to increase the accuracy of dates entered in the date of birth question. Findings from user research testing (2017:2, 2018:2, 2019:1, 2020:2) suggested the additional page to confirm respondents age helped, and made some respondents go back to the date of birth question to correct the digits entered.

We recommend using the following confirmation of age question.

You are XX years old. Is this correct?

[ ]    Yes, I am XX years old
[ ]    No, I need to change my date of birth

In the White Paper Help Shape our Future: The 2021 Census of Population and Housing in England and Wales (PDF, 967KB), we committed to continuing to ask questions on marital status and civil partnerships, and household relationships.

Marriage of same-sex couples has been legalised since the last census, and in accordance with the undertaking given in the topic consultation report, the Office for National Statistics (ONS) conducted testing to review the marital status question and response options to ensure all the necessary data are collected, and will seek to ensure that all legally recognised statuses are covered.

This section provides the details on how the questions on marital and civil partnership status, and relationship matrix evolved from the beginning of testing through to the final recommended questions for Census 2021.

The starting point for testing the questions on marital status and household relationships was the 2011 Census question designs. You can see these questions on the 2011 Census Household Questionnaire (PDF 1.8 MB). However, online, a very different approach was needed to collect relationships in a way that aligned with Government Digital Service’s Service Standard.

Marital and civil partnership status

Marital or legal partnership status is used by a wide range of users, including central government, to monitor the impact of government policy and service delivery. Marriages and civil partnerships are protected characteristics under the Equality Act 2010. Marriages and civil partnership legislation has changed since the 2011 Census. The two new legislations, Marriages (Same Sex Couples) Act, 2013 and the Civil Partnerships, Marriages and Deaths (Registration Etc.) Act 2019 include same-sex marriages and opposite-sex civil partnerships.

2011 Census question

On 27 March 2011, what is your legal marital or same-sex civil partnership status?

• Never married and never registered a same-sex civil partnership
• Married
• In a registered same-sex civil partnership
• Separated, but still legally married
• Separated, but still legally in a same-sex civil partnership
• Divorced
• Formerly in a same-sex civil partnership which is now legally dissolved
• Widowed
• Surviving partner from a same-sex civil partnership

On the 2011 Census paper questionnaire marriage response options were on the left-hand side, and civil partnership response options were on the right-hand side of a column.

Legal status changes

Prior to the 2011 Census, civil partnerships were legalised (via the Civil Partnerships Act, 2004), which allowed same-sex couples (only) to enter in to a civil partnership. Marriages and civil partnerships are protected characteristics under the Equality Act 2010. Since the 2011 Census, the Marriages (Same Sex Couples) Act, 2013 was introduced, which legalised same-sex marriages. As a result of this, new response categories were introduced to the 2011 Census marital and civil partnership status question to ensure the collection of data on registered civil partnerships.

In 2015, we committed to carry out further research to reviewing the marital and civil partnership status response categories and question design to reflect the changes introduced by the Marriage (Same Sex Couples) Act 2013 that enabled same-sex couples to marry. As part of this we undertook a programme of testing, evaluation and consultation to identify ways of collecting data on marriages and civil partnerships to meet user need. Census 2021 will therefore be the first opportunity to collect data on same-sex marriages.

In 2019, the Civil Partnerships, Marriages and Deaths (Registration Etc.) Act 2019 became law, meaning that opposite-sex couples can also enter into civil partnerships. Our development and testing activities included provision for potential collection of opposite-sex civil partnerships to prepare for the possibility that they would become legal prior to Census 2021. Census 2021 marital status question has been amended and will therefore also collect information on opposite-sex civil partnerships.

For Census 2021, a new two-stage question design was tested to best capture marital and civil partnership legal statuses. The first stage is used to collect legal status. The second stage asks if the respondent is, or was, in a marriage or civil partnership with someone of the opposite or same sex.

Developing a two-stage marital and civil partnership status question

To begin our testing, we developed a two-stage question based on the design of the 2011 question. This two-stage design allowed the collection data on marriages between partners of the opposite-sex and partners of the same-sex. The question design asked: “On [date], what is your legal marital or same-sex civil partnership status?” with the following response options:

• Never married and never registered a same-sex civil partnership
• Married or in a registered same-sex civil partnership
• Separated, but still legally married or in a same-sex civil partnership
• Divorced or dissolved same-sex civil partnership
• Widowed or surviving same-sex civil partner

The second stage of the initial question design tested was “What type of partnership is (was) this?” with these response options:

• A marriage to a partner of the opposite sex
• A marriage to a partner of the same sex
• A same-sex civil partnership

We investigated public acceptability of the redesigned marriages and civil partnerships question by conducting an omnibus survey (2017:19). The face-to-face interviews revealed that most participants found the redesigned marital and civil partnership status question acceptable. In total, 92% of respondents would answer the redesigned question accurately and continue completing the census. Only 5% of respondents found the redesigned question unacceptable. However, 82% of those would still answer the question accurately and stated that they would continue to complete the census. The “Never married and never registered a same-sex civil partnership” option was found to be the least acceptable response option, possibly because respondents were looking for “single” or an option for cohabiting partners.

Based on the overall positive findings of the public acceptability testing, it was recommended that the redesigned question was cognitively tested to assess how respondents understood the question and if they were able to respond accurately.

The first round of cognitive testing (2017:23) used the same two-stage design that was tested during the first quantitative test (2017:19). Participants answered the questions and were then asked about their experience by a trained interviewer. The results were that overall, participants found the questions acceptable and that the questions were easy to answer.

The recommendations from this round of testing were:

• in the first stage of the question, simplify the questions as some participants thought the text of the question and response options too long and complicated; one suggestion was to separate marriage and civil partnerships
• in the first stage of the question, remove “same-sex” from the civil partnership question as some participants thought the question was only asking about same-sex relationships
• include guidance or details to indicate that the question is asking about opposite-sex as well as same-sex legal relationships
• provide guidance on the definition of “civil partnerships”

After this round of cognitive testing we conducted informal interviews at two LGBT+ events in Brighton and Winchester (2018:4). During this round of testing three question designs, all with two stages, were shown to participants:

Version 1: the version used in the previous round of cognitive testing (2017:23)

Version 2: a version with the same first stage as Version 1 but asking “Is or was this:” with response options “an opposite-sex marriage”, “a same-sex marriage” and “a same-sex civil partnership” in the second stage

Version 3: a similar stage one question as the first two versions but with all the response options separated out and shown in two columns (as per the 2011 Census paper design). For example, the option “Married or in a same-sex civil partnership” was separated into response options “Married” and “In a same-sex civil partnership” and these were shown in separate columns on the same line. The second stage for this question asked, “Is or was this relationship:”, with response options “between opposite-sex partners” or “between same-sex partners”.

The three versions of the questions were shown in randomised order to the participants at the two events. The participants were asked to give their comments on each version by thinking aloud as they read them.

The results showed that participants preferred the first version, which was the version used in the previous round of cognitive testing (2017:23) that used a two-stage design. Participants did not like the layout of Version 3, which showed the response options in two columns. Some participants misunderstood civil partnerships as cohabitating relationships between those of opposite sex rather than a legally recognised partnership between same-sex couples. Some participants also commented that it was unclear how people from the trans community should answer the questions as the questions referred to sex and not gender.

Recommendations from this round of testing included:

• provide guidance for members of the trans community to help them answer this question and understand why certain terminology has been used (for example the use of “sex” instead of “gender”)
• keep response options as a one-column list
• provide information about civil partnerships
• take forward the wording of Version 1, as it was the easiest to understand and considered acceptable by the greatest number of participants
• consider separating out the option in the first stage question for clarity (for example using “Married” and “In a same sex civil partnership” instead of a combined option)

Ordering of marital and civil partnership status response options

As participants in the previous rounds of testing had expressed a preference for separating out the response options in the first stage of the question, further testing (2018:19) was carried out to investigate the preferred order of response options. An online survey was used to test out two versions of the first stage of the question that asked “On [date], what was your legal marital or same-sex civil partnership status?”:

Version A listed the options in this order:

• Never married and never registered a same-sex civil partnership
• Married
• Separated, but still legally married
• Divorced
• Widowed
• In a registered same-sex civil partnership
• Separated, but still legally in a same-sex civil partnership
• Formerly in a same-sex civil partnership which is now dissolved
• Surviving partner from a same-sex civil partnership

Version B listed the options in this order:

• Never married and never registered a same-sex civil partnership
• Married
• In a registered same-sex civil partnership
• Separated, but still legally married
• Separated, but still legally in a same-sex civil partnership
• Divorced
• Formerly in a same-sex civil partnership which is now dissolved
• Widowed
• Surviving partners from a same-sex civil partnership

Respondents to the online survey were asked to answer both versions of the question and then were asked to select which version they preferred and to explain the reasons for their preference. Participants were recruited via the ONS twitter account and the GovDelivery and census distribution lists. Of all responses received over half of the participants (57%) were married.

Version B was preferred to Version A by a statistically significant greater number of participants. The most common explanations for the preference was that Version B was more inclusive, showed equality between marriage and civil partnership, had a more logical order and was clearer and less confusing than Version A.

The recommendation from this round of testing was to show the response options in the order used in Version B.

Developing a three-stage marital and civil partnership status question

For the second round of cognitive testing (2018:22), two new versions of the question were used and both contained three stages. A three-stage design was tested to assess how well a design would work that removed those who had never been married or in a registered civil partnership through using a short question and then reducing the range of options in the second stage.

Version A asked:

Have you ever been legally married or in a registered civil partnership?

• Yes
• No

On [date], what is your legal marital or registered civil partnership status?

• Married
• In a registered civil partnership
• Separated, but still legally married
• Separated, but still legally in a civil partnership
• Divorced
• Formerly in a civil partnership which is now legally dissolved
• Widowed
• Surviving partner from a civil partnership

What type of partnership is (was) this?

• Marriage to a partner of the opposite sex
• Marriage to a partner of the same sex

Version B had the same questions and response options but used “same-sex civil partnership” instead of “civil partnership” and “registered civil partnership” in the second stage. Respondents were routed out of the questions if they answered “No” to the first stage, or if they indicated they were in a civil partnership in the second stage.

These versions made changes suggested by the previous rounds of testing:

• simplified answering for those who had never been married or in a civil partnership
• listed marital and civil partnerships statuses in separate response options
• Version A removed the term “same-sex” which had caused some confusion in previous testing

Participants answered census questions, including Version A of the marital and civil partnership status questions, and were then interviewed about their experience by trained interviewers. During the interview, they were shown Version B of the questions and asked for their comments. A paper questionnaire was available, however none of the participants asked to use it, so all participants completed the questions on an online questionnaire. Participants were chosen to be a representative sample of different sexual orientations and marital statuses, including those in civil partnerships and those that had never been married or in a civil partnership.

The results showed that the question and response options were generally well understood and found to be acceptable by the participants. The inclusion of civil partnerships and same-sex marriages was especially welcomed.

Participants thought that the phrase “same-sex” was not needed when talking about civil partnerships but preferred the use of “registered civil partnership” as they thought this made civil partnerships seem equal to marriages. When the term “registered” was removed from civil partnership, participants were more likely to be confused and unsure whether to count informal co-habiting relationships as civil partnerships.

In general, the participants felt that the questions on marital and civil partnership status should have two stages, rather than three. However, while some participants thought that while three stages were a lot, they also understood the need to split the question in this way.

The recommendations from this testing were:

• use “registered civil partnership” instead of “same-sex civil partnership”
• investigate if a two-stage or three-stage question is preferred by respondents

Further testing to determine the preference for a two-stage or three-stage design

We carried out two rounds of testing: informal interviews at two Pride events in England and Wales (2018:31) and an internal focus group (2018:36). The purpose of these was to:

• investigate any preference for a two- or three-stage question
• investigate different wordings of the question
• design a question that reflected the possibility of civil partnerships being extended to opposite-sex couples in future, because of the Civil Partnerships, Marriages and Deaths (Registration, Etc.) Act 2019, which had been introduced at a late stage in the question development process

At the Pride events, participants were shown three different question designs, all of which had been used in previous rounds of testing. The differences between the three versions were:

Version 1 (three-stage design): the first question asked: “Are you currently, or have you ever been legally married or in a registered civil partnership?”, the second question asked: “On [date], what is your legal marital or registered civil partnership status?”; and the third question asked: “What type of partnership is (was) this?” with response options “Marriage to a partner of the opposite sex” and “Marriage to a partner of the same sex”.

Version 2 (two-stage design): This version omitted the first question asked in the three-stage design and the final question asked: “What type of partnerships is (was) this?” with response options “Marriage to a partner of the opposite sex” and “Marriage to a partner of the same sex”.

Version 3 (two-stage design): This version again omitted the first question asked in the three-stage design and the final question asked: “Who are (were) you in a marriage with?” with response options “A partner of the opposite sex” and “A partner of the same sex”.

At the focus group, participants were shown two of the question designs that had been used at the Pride event: the two-stage question (Version 2) and the three-stage question (Version 1).

At both events, participants were shown the versions of the questions and asked for their comments. The findings from these rounds of testing were that participants generally favoured the two-stage question designs and found it acceptable and easy to answer in this format. Many participants commented that information about why the questions were being asked should be included in the guidance.

In the focus group, participants agreed that the final question should include the terms marriage and civil partnership in the question stem. For example, one group suggested that the question should be “What type of legal marriage or registered civil partnership is (was) this?”, with response options “To a partner of the opposite sex” and “To a partner of the same sex”. Changing the question this way would also ensure that it would capture accurate information if, as expected, civil partnerships are extended to opposite-sex couples by the time the census takes place in 2021.

The main recommendations from these two rounds of testing were:

• the question should be a two-stage question
• the second stage should include reference to marriages and civil partnerships for both opposite-sex and same-sex couples
• guidance for these questions should include information about why these questions are being asked

Acceptability testing of the two-stage marital and civil partnership status question

The two-stage question was taken forward into quantitative testing (2018:41). Participants took part in an online survey that included a wide range of census questions such as those on qualifications, ethnicity, employment and health. The questionnaire also included the following two-stage question on marital and civil partnership status:

• On [date] what is your legal marital or registered civil partnership status?
• Who is (was) your legal marriage to?

The second stage was shown only to those who had reported they were married, divorced, separated or widowed in the first stage. The second stage response options were “Someone of the opposite sex” and “Someone of the same sex”. After the census questions, participants were asked about the acceptability of the census questions they had just answered.

The purpose of this testing was to find out how acceptable the second stage question was to respondents, particularly the acceptability of the response options “Someone of the opposite sex” and “Someone of the same sex”.

The results from this test showed:

• in total, 75% of the respondents found the second stage of the marital and civil partnership status question acceptable, 11% found it to be unacceptable
• overall, 92% of participants said they would answer the second stage of the marital and civil partnership status question accurately and complete the rest of the questionnaire and only 1% said it would cause them to stop answering any further questions in the census
• participants who were married to someone of the opposite sex were more likely to find the second stage question acceptable than those in a same-sex marriage (although it should be noted that in the sample only 2% of participants were in a same-sex marriage)

We recommended the second-stage question wording as tested for Census 2021 as shown in the section Questions recommended for Census 2021.

Household relationships

2011 Census household relationship matrix

The 2011 Census collected all relationships between Person 1 to Person 6, and the relationships of Person 7 through Person 13 to Person 1, and each other in the Household Continuation questionnaire. Each household and household continuation questionnaire collected up to six persons, it did not capture the relationships for example, between Person 8 to Person 2 as this relationship would be across two forms. Names were collected for each person. The options available in the matrix were:

• Husband or wife
• Same-sex civil partner
• Partner
• Son or daughter
• Step-child
• Brother or sister
• Step-brother or step-sister
• Mother or father
• Step-mother or step-father
• Grandchild
• Grandparent
• Relation – Other
• Unrelated (including foster child)

The order of the response options remains the same as in the 2011 Census. The response option “Same-sex civil partner” has been updated to “Legally registered civil partner” to reflect that civil partnerships are also available to opposite-sex couples.

The Census 2021 household and household continuation paper questionnaire remains identical to the 2011 Census, except it collects up to five persons per questionnaire; this was a mitigation for space constraints on the paper questionnaire.

Household relationship question for the online questionnaire

The household relationships question asks how members of the household are related to each other. Although the 2011 question on paper had no negative feedback or requests to make a change, we needed to consider the ability of respondents to answer this question online as well as on paper in Census 2021.

Online, the household relationships question will use the names of those who live in the household (collected earlier in the form) to help respondents answer the question. For example, if a householder with the name “Jade Smith” has been added, then the household relationships question will be “Jade Smith is your…” with the list of response options. Once a response option is selected, the text at the top and bottom of the page updates to reflect the relationship. For example, the text might say at the top and bottom of the page “Jade Smith is your husband or wife”. The reason for repeating the text at the bottom of the page is because of the length of the list, which means that on mobile devices the text at the top cannot be seen when tapping the ‘Save and continue’ button at the bottom of the list. Therefore, repeating the text at the bottom ensured that respondents can confirm the household relationship before continuing onto the next page.

For relationships between householders, the text will, for example, say “Thinking of Jade Smith, Barry Smith is their…”. Once a relationship is selected the text is updated. For example, the text might say “Thinking of Jade Smith, Barry Smith is their son or daughter”.

The number of people who can be recorded on the online household relationship question is limitless. Respondents are advised when they add the 16th person in the household to phone the customer helpline to be assessed and to ensure that they are one household and not a communal establishment.

Census 2021 question

[Name] is your [Relationship selected]
Complete the sentence by selecting the appropriate relationship.

[ ]    Husband or wife
[ ]    Legally registered civil partner
[ ]    Partner
[ ]    Son or daughter
[ ]    Stepchild
[ ]    Brother or sister (including half-brother or half-sister)
[ ]    Stepbrother or stepsister
[ ]    Mother or father
[ ]    Stepmother or stepfather
[ ]    Grandchild
[ ]    Grandparent
[ ]    Other relation
[ ]    Unrelated (including foster child)

[Name] is your [Relationship selected]

Development of the order for household relationship response options

We designed three card sorting activities (2017:13; 2017:20) in order to determine the most appropriate order and whether there was any preference of the response categories within the household relationships question. We also wanted to establish how to best present the order in an online and paper format because of the long list of response options.

The three activities involved:

Firstly, a group of people that worked for the ONS, sorting paper cards. Secondly, a group of people recruited via ONS social media accounts, taking part in an online card sort.

Lastly, students from two universities taking part in paper card sorting that we selected purposively for the study as students are more likely than the general population to live in households with unrelated members. University students were approached as, in addition to asking for general feedback, that the “unrelated” response option was at the bottom of the 2011 list and we wanted to find out where the students thought it should be placed. We wanted to assess if we could identify any issue with the order of the list when answering the question online considering that it is a long list to navigate when using a mobile phone.

Preference for double- or single-banked

We wanted to establish if the participants preferred a single column or double-banked design and their preference of response option order when using a double-banked design. In each activity, participants were told about the purpose of the activity, given access to the list of household relationships (either via a set of paper cards or an online list) and asked to place them in the order they felt most appropriate for the question asked on the census.

They were first asked to place them into a single column of responses, then asked to perform the sorting again but sort the responses into two columns. A two column (or double-banked) variant was included in the testing to assess the optimal order suggested because the final format for the online questionnaire (single- or double-column format) was still to be confirmed at the start of the testing. Online, the participant saw the initial list of relationships in a random order. The paper cards were shuffled between use to change the order.

The results from the three rounds of testing were collated to create one recommended order of relationships.

Testing household relationships in two columns

When the question was sorted into two columns, no clear consensus on the recommended order of responses could be established. A two-column layout may require respondents to scroll horizontally on smaller devices such as smartphones or could result in a change meaning that the layout adapts to put all the options into one column. Either outcome would be different from the experience of those completing the questionnaire on paper. Feedback from participants also suggested a one-column layout is preferable despite it being a much longer list. For this reason and because of potential mode effects, the decision was made to use a one-column layout for both electronic and paper questionnaires.

Testing household relationships in a single column

The recommended order of the household relationships response options was:

• Husband or wife
• Partner
• Same-sex civil partner
• Son or daughter
• Stepchild
• Mother or father
• Stepmother or stepfather
• Brother or sister
• Stepbrother or stepsister
• Grandparent
• Grandchild
• Relation – other
• Unrelated (including foster child)

The differences from the order recommended by participants in the three rounds of testing and the order used in the 2011 Census were:

• “partner” swapped positions with “same-sex civil partner”
• the mother and father options (including step-parent options) swapped positions with the brother and sister (including step-siblings) options
• “grandparent” swapped position with “grandchild”

Although the card sorting exercises suggested that “Partner” should come before “Same-sex civil partner” in the list, there were some concerns that this could lead to respondents choosing “Partner” instead of “Same-sex civil partner” because it came first in the list and respondents tended to pick the first appropriate response option. Because of these concerns about data quality, it was decided to use the same order as suggested in the 2011 Census for these options, putting the “Same-sex civil partner” option before the “Partner” option.

One change was made to the wording of the question; because of the potential for civil partnerships to be extended to opposite-sex couples before the Census 2021, the option “Same-sex civil partner” was changed to “Legally registered civil partner”. This wording was tested during development of the marital and civil partnership status question (2018:41) and was found to be well understood and acceptable to respondents.

There were no data concern issues regarding the order of the parent and sibling options, or the grandparent and grandchild options that had been used previously, so the recommendation was to keep these in the same order as the 2011 Census.

In 2011 Census respondents were advised if they have half-siblings to select the response option “Brother or sister”. Findings from focus groups (2017:20) suggests this guidance should be clearer. For Census 2021 on the online questionnaire, this has been included as an instruction under the response option “Brother or sister” for clarity.

Therefore, the final ordered list of recommended response options for Census 2021 is:

• Husband or wife
• Legally registered civil partner
• Partner
• Son or daughter
• Stepchild
• Brother or sister (including half-brother or half-sister)
• Stepbrother or stepsister
• Mother or father
• Stepmother or stepfather
• Grandchild --Grandparent
• Other relation
• Unrelated (including foster child)

Welsh language question development

Between 2017 and 2018, an external agency with Welsh speaking researchers was commissioned to undertake focus groups (2017:17) and a series of cognitive interviews (2017:18). In 2018, further cognitive interviews (2018:40) were undertaken by the same agency. 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 adhere 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. This group includes Welsh language and policy experts from the Welsh Language Commissioner and the Welsh Government and was convened to give advice on the accuracy, clarity and acceptability of the language as well as other policy issues pertaining to the Welsh language and bilingual design.

The findings of this research led to the following recommendations for the Welsh language question design:

For the date of birth question, participants suggested the translation “Beth yw eich dyddiad geni?” should be used for “What is your date of birth?”

For the marital status and civil partnership question, we recommend using the term “priodas ag un o rhyw gwahanol” should be used for “opposite sex marriage”; the term “gymar” (partner) was preferred in relation to someone who is deceased. In relation to the term for “never married and never registered a same-sex civil partnership”, participants suggested the use of “erioid” for “never”.

The design of the questions recommended for Census 2021 was informed by the research and testing detailed in this report. These have now been finalised through the census secondary legislation.

The images that follow show the questions on the electronic questionnaire. We have also published the paper questionnaires for Census 2021.

Guidance text and instructions are not part of the legislation, but we consider these to be finalised as well. However, it is possible that guidance text or instructions may change if there is enough evidence to support doing so.

Figure 1: Online question on date of birth in English and Welsh

England and Wales

Source: Office for National Statistics

Download this image Figure 1: Online question on date of birth in English and Welsh

.png (50.3 kB)

Figure 2: Online question confirming age in English and Welsh

England and Wales

Source: Office for National Statistics

Download this image Figure 2: Online question confirming age in English and Welsh

.png (90.6 kB)

Figure 3: First online question on marital and civil partnership status in English and Welsh

England and Wales

Source: Office for National Statistics

Download this image Figure 3: First online question on marital and civil partnership status in English and Welsh

.png (279.1 kB)

Figure 4: Second online question on marital and civil partnership status in English and Welsh, if currently married England and Wales

England and Wales

Source: Office for National Statistics

Download this image Figure 4: Second online question on marital and civil partnership status in English and Welsh, if currently married England and Wales

.png (48.3 kB)

If on the first-stage question (Figure 3) respondents select “Separated, but still legally married”, “Divorced”, or “Widowed”, the following question stem changes to “Who was your legal marriage to?”

Figure 5: Second online question on marital and civil partnership status in English and Welsh, if currently in civil partnership England and Wales

England and Wales

Source: Office for National Statistics

Download this image Figure 5: Second online question on marital and civil partnership status in English and Welsh, if currently in civil partnership England and Wales

.png (54.8 kB)

If on the first-stage question (Figure 3) respondents select “Separated, but still in a registered civil partnership”, “Formerly in a civil partnership which is now legally dissolved”, or “Surviving partner from a registered civil partnership”, the following question stem changes to “Who was your registered civil partnership to?”

Figure 6: Online question on household relationships to Person 1 (where Person one is completing the questionnaire), in English and Welsh

England and Wales

Source: Office for National Statistics

Download this image Figure 6: Online question on household relationships to Person 1 (where Person one is completing the questionnaire), in English and Welsh

.png (254.5 kB)

Figure 7: Online question on household relationships between members of household other than the person completing the questionnaire, in English and Welsh

England and Wales

Source: Office for National Statistics

Download this image Figure 7: Online question on household relationships between members of household other than the person completing the questionnaire, in English and Welsh

.png (313.1 kB)

Question order

Date of birth and Marital and civil partnership status

In the previous census in 2011, the question on sex came before the question asking date of birth. For Census 2021, a voluntary question on gender identity has been added; this is only asked of those aged 16 years or older. A guidance note has been added to the sex question saying that a question on gender identity will follow, however this is only applicable to those aged 16 years or older and should not be shown to respondents below the age of 16 years online (it will appear on the paper questionnaire). Therefore, the question on date of birth has been moved so that it comes before the sex question, so that age is known before the sex question has been asked and the age-related guidance note can be shown online only if necessary.

Household relationships

Cognitive testing (2018:33) was carried out with participants who were purposively recruited to represent complex household compositions. The testing included three versions of the questions asking who usually lives at the address and household relationships: two versions asked participants to record everyone who usually lives at the address followed by questions asking about all the relationships between these people, and one version asked participants to record each person who lives at the address followed by the household relationships for that person, and then record the next person and their relationships and so on, until all people and relationships had been recorded.

Results showed that integrating the household relationships question into the questions on who usually lives at the address increased the cognitive burden on participants and could affect the accuracy of data collected. The recommendation was to keep the household relationships question separate. Therefore, respondents are asked to record all household relationships after they have recorded all the people who usually live at the address.

Question evaluation

The question designs put forward in this report are based on extensive research and assessment using evaluation criteria that were set out in the publication The 2021 Census – Assessment of initial user requirements on content for England and Wales: Response to consultation (PDF, 796KB).

The evaluation considered the potential impact that including a topic on the census would have on data quality, public acceptability, respondent burden, financial concerns, and questionnaire mode. The evaluations were used in conjunction with the user requirements criteria to steer the development of the census questions and questionnaire.

A topic that has been assessed as having a “High” potential for impact is closer to the threshold for exclusion from the census than a topic that has been assessed as having a “Low” potential for impact.

After completing the research and development phase, we evaluated the recommended questions against the same criteria using an updated tool that considers the type of evidence we have available and the Census 2021 context. A description of this updated evaluation tool is provided in the Question and questionnaire development overview for Census 2021.

All questions meet our thresholds to ensure reliable information will be collected in Census 2021.

Table 2 provides the updated evaluation scores for the questions on demography. We present the evidence used to assess questions as having a “Medium” or “High” potential for impact on the evaluation criteria in Annex 3.

Question harmonisation

As in previous censuses, there will be separate censuses in Scotland and Northern Ireland. The questions for England and Wales have been developed through close collaboration with National Records of Scotland (NRS) and the Northern Ireland Statistics and Research Agency (NISRA), who are responsible for conducting the censuses in Scotland and Northern Ireland, respectively.

For the date of birth question, NISRA have indicated that they plan to use DD MM YYYY within the text boxes on the paper form (the Office for National Statistics (ONS) and Scotland’s statistics office are using the terms “Day”, “Month” and “Year” written above the corresponding boxes). For Scotland, the online questionnaire will instruct users to ‘use the format: DD MM YYYY’ and include these within the text boxes.

For the marital and civil partnership status question, the NRS are using the same first question of the two-stage design, aside from using the response option “Surviving partner from a civil partnership”. For Scotland, the date in the question will reflect the 20 March 2022 Census date. NISRA are not including the date in the question stem and are including the response option “single” in their census.

For the household relationship question, NISRA are including the same question and response options as the ONS in their censuses.

While Scotland’s question design is similar to the ONS and NRS, there are some differences for Scotland:

• they do not include “If members are not related…” in Scotland's question
• they have “first name(s)” rather than “first name”
• they have, for example, “Relationship of Person 2 to Person:” rather than “How is Person 2 related to Person:”
• they have the response option “Registered civil partner” (not “Legally registered civil partner”)
• they do not include “including half-brother or half-sister” in the brother or sister option, this messaging is elsewhere
• they have the response option “Other relation (including in-laws)” rather than “Relation – other”

We recognise that each statistical office has its own user and respondent needs; however, we aim for harmonisation of census questions and topics where possible to produce UK-wide statistics that are consistent and comparable. The demography questions were developed for use in the context of Census 2021 in England and Wales. Therefore, it is possible that in different circumstances a different approach may be more suitable. For example, some social surveys use different questions such as the government statistical service harmonised principles if they are found to be more appropriate in a different context.

In the 2021 Census topic consultation response, the Office for National Statistics (ONS) made clear commitments to the public. We committed to continuing to ask the questions on date of birth, marital or civil partnerships status and household relationships. We also made the following commitments:

Commitment to ensure the marital and civil partnership status questions were kept up-to-date and in line with current legislation

We updated the marital and civil partnership status questions so that they would reflect possible changes to the law that allow same-sex couples to marry and opposite-sex couples to register a civil partnership introduced by the Marriages (Same Sex Couples) Act, 2013 and Civil Partnerships, Marriages and Deaths (Registration Etc.) Act 2019.

Improve the respondent experience, reduce the respondent burden and investigate ways to improve data quality

We met these commitments by extensively testing the household relationships question to ensure the question is designed to minimise respondent burden. We were helping respondents to enter household relationships more easily online by using names instead of placeholders (for example, we used “Thinking of John Smith…” instead of “Thinking of Person 1…”) to help them enter the household relationships more easily.

We ensured the marital and civil partnerships status questions were considered acceptable by a large majority of respondents, and that they did not result in an unacceptable number of dropouts from the questionnaire.

We tested various possible layouts of the date of birth question for the online questionnaire to find the design that would give us the best data quality with the least respondent burden. We then used the date of birth data entered online to calculate age and therefore enabling respondents to check their answers.

References to tests take the form (Year: Test number). “Year” refers to the calendar year the test was undertaken in and the test number is the position of the test within the year considering all testing that took place in that year. For example, the fifth test conducted in 2017 would be referenced as (2017:5).

A full description of each of these items can be found in summary of testing for Census 2021.

Evaluation of date of birth question

The potential impact for the date of birth question on data quality, public acceptability, respondent burden and financial concerns were assessed as “low”.

Potential for impact on questionnaire mode: “Medium”

The online question does not allow respondents to proceed without having answered the date of birth question. Furthermore, the online question is broken into two sections, first asking for date of birth and then asking the respondent to confirm their age calculated from the information given in the date of birth question. These differences between paper form and electronic questionnaire led to the potential impact on questionnaire mode being assessed as “medium”.

Evaluation of marital and civil partnership status question

The potential impact for the marital and civil partnership status question on data quality, public acceptability, respondent burden and financial concerns were assessed as “low”.

Potential for impact on questionnaire mode: “Medium”

The question on the electronic questionnaire has radio buttons and therefore prevents multi-ticking, however, more than 1% of respondents select multiple response options on the paper form. These differences between paper form and electronic questionnaire led to the potential impact on questionnaire mode being assessed as “medium”.

Evaluation of married or in a registered civil partnership with question

The potential impact for the married or in a registered civil partnership with question on data quality, public acceptability, respondent burden and financial concerns were assessed as “low”.

Potential for impact on questionnaire mode: “Medium”

The second stage of the marital and civil partnership status question asks again for sensitive information with the potential of respondents choosing a more socially acceptable response option, social desirability answers reduce when answering online. The question displayed on the electronic questionnaire depends on the response option chosen in the first stage of the marital and civil partnership status question and therefore looks different. In addition, the question on the electronic questionnaire has radio buttons and therefore prevents multi-ticking. All of these differences between paper form and electronic questionnaire led to the potential impact on questionnaire mode being assessed as “medium”.

Evaluation of household and family relationships question

The potential for impact on data quality, public acceptability and financial concerns have been assessed as “low”.

Potential for impact on respondent burden: “Medium”

On the paper questionnaire, the respondent must restate the names of household members, which is used to help respondent understanding of the question. On the electronic questionnaire, the names of other household members are piped into the question stem, however some respondents did not understand the question or the type of answer that was expected. There is also a lot of repetition in the question stem and responses. Therefore, the potential for impact on respondent burden has been assessed as “medium”.

Potential for impact on questionnaire mode: “Medium”

During testing (2018:42), there were more errors on the electronic questionnaire because of respondents misunderstanding or not reading the question stem. Once these errors were identified by the respondent, they would return to review the first stage of the multi-stage question. The paper questionnaire has a name write-in, while the electronic questionnaire will pipe in names to the question stem automatically. The two modes have different instructions and are visually different from one another. Therefore, the potential for impact on questionnaire mode has been assessed as “medium”.