Sex and gender identity 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. This outlined the Office for National Statistics’s (ONS’s) proposal to collect information on gender identity, in addition to the existing question on sex. This would meet the user need for better quality information for equality monitoring and to plan and provide services.

There are currently no official figures for those who identify their gender as different from the sex registered at birth.

ONS research and consultation showed a clear need for information on gender identity, to support work on policy development and service provision and to further equality, including under the relevant equality law.

Recommended sex question for Census 2021

What is your sex?
A question about gender identity will follow later on in the questionnaire
[ ] Female
[ ] Male

This question wording and response options are unchanged from the 2011 Census. We will continue to collect this data in a way that is consistent with previous censuses.

Recommended gender identity question for Census 2021

Is the gender you identify with the same as your sex registered at birth?
This question is voluntary
[ ] Yes
[ ] No
(Enter gender identity)

The gender identity question is voluntary. It will only be asked to respondents aged 16 years and over.

Since the publication of the White Paper, the Office for National Statistics (ONS) has conducted and concluded the final phase of testing on the topic of gender identity. This report provides links to previously published research and the findings of additional testing that led to the final recommended questions for Census 2021 for 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 as well as Census Regulations for England and for Wales.

As the topic of gender identity is closely related to the topic of sex, the two topics are discussed together in this report.

The evidence base for the recommendations made in the White Paper is discussed in Section 5: Research that led to the 2018 White Paper recommendations. The evidence base for the finalisation of the questions for Census 2021 is discussed in Section 6: 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, the ONS published its response to the Census 2021 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

This response included a commitment to continue to collect data on sex and to further investigate whether to collect data on gender identity and how this could be achieved.

The topic consultation revealed a clear requirement for information on gender identity for policy development and service planning, in particular with regard to the provision of health services. Stakeholders also told us that the data was required for monitoring equality.

We received feedback from members of the public who reported that they were unable to complete the 2011 Census accurately as it included the current Government Statistical Service (GSS) harmonised sex question, which only has two categories: “male” and “female”.

A detailed summary of the consultation responses relating to the topic of gender identity can be found in the gender identity topic report (PDF, 728KB). As information about gender identity had not previously been collected in the census, nor on any ONS social surveys, we published a gender identity research plan (PDF, 799KB) at the same time as our consultation response. This research plan informed our position on this topic and detailed our proposed plan of work. Within these reports, the ONS made clear commitments to the public. We have provided an update on how we met these commitments in Annex 1 of this report.

With the topic being a new area of work for the ONS, we also launched the gender identity pages on the ONS website to provide updates of our work.

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 gender identity 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 those who identify their gender as different from their sex registered at birth as well as those who identify their gender as being the same as their sex registered at birth.

Our initial work on this topic focused on user engagement to better understand user needs and to develop a clearer understanding of the different concepts. We held a gender identity workshop in August 2016 and a gender identity update event in June 2017 with data users. This stakeholder engagement supported a consistent data need for a count of the trans population, including individuals of all ages, and all people who identify their gender as different from their sex registered at birth. The engagement also emphasised that respondents must be able to self-identify how they wish.

In August 2017, we published a gender identity update to the Trans Data Position Paper (PDF, 186KB) published in 2009. This review outlined developments around the topic of gender identity including:

• legislation
• the Women and Equalities Committee Transgender Equality inquiry
• data collection and question development worldwide
• details of our research, testing and findings up to that date

In December 2017, we published a further census topic research update, which summarised our exploratory research on question design and progress on understanding the concepts surrounding the topic of gender identity. A major focus of our research was ensuring there was no risk to the quality of the data collected on a person’s sex. Sex is one of the most frequently used and important characteristics the census collects. Like gender reassignment, “sex” is also a protected characteristic, as set out in the Equality Act 2010. We concluded that none of our approaches to a gender identity question to that point would fully meet user needs. We therefore committed to undertaking further question development and testing, before recommending that gender identity was included as a topic in the census.

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. This outlined our proposals to collect information on gender identity, in addition to the current question on sex, to meet the needs for better quality information for equality monitoring and to provide the information that users need to plan and provide services.

Alongside the White Paper, we published a further census topic research update. This provided additional details of the research that supported the gender identity recommendations announced in the White Paper. At this time, we were still analysing the findings of a large-scale quantitative test to inform the final question design for the gender identity topic. This report provides the findings from this final testing and shows how we finalised the gender identity question design recommended for Census 2021.

We have since published an updated position on the potential to use other data sources, Exploring existing data for gender identity and sexual orientation, including administrative data sources, on the topic of gender identity. This research shows that there are currently no administrative data sources that can provide a reliable estimate of the population who identify their gender as different to their sex registered at birth.

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

Throughout this report, we use the widely used word “trans” to describe all those whose gender is not the same as the sex they were registered at birth. Language around this topic is still emerging. Our use of the term is inclusive of a range of genders, including:

• binary male or female genders when not the same as registered at birth
• non-binary genders such as those on a continuum between male and female
• non-gendered identities (neither male nor female)

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

As set out in the White Paper Help Shape our Future: The 2021 Census of Population and Housing in England and Wales, with further details provided in the December 2017 and December 2018 census topic research updates, our testing on the gender identity topic involved investigation of four different approaches, before a final question design was recommended. These four approaches were:

• Approach 1: exploratory research – a non-binary sex question
• Approach 2: exploratory research – a sex question and a gender question
• Approach 3: development – a gender question and a gender identity question
• Approach 4: development – a sex question with guidance that a gender identity question will follow and a gender identity question

Question design was complex because the language related to the topic of gender identity is continually developing and there are many identities with which people may want to identify.

The first two approaches were exploratory and conducted in parallel. The third approach moved us towards a method for collecting data that met our definition of the trans population and allowed us to consider the impact of the tested designs on the collection of high-quality data on sex. The fourth and final approach proved more viable in designing a question for Census 2021. The initial findings from the testing on Approach 4 allowed us to put forward the recommendations made in the White Paper and were discussed in the December 2018 census topic update.

We provide a full list of the tests used in the development of the sex and gender identity topics in Annex 2, and further details are in the summary of testing for Census 2021.

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).

The tests utilised a range of different 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. All testing included respondents from a wide range of backgrounds. This included those who identify their gender as being the same as their sex registered at birth as well as those who identify their gender as being different to their sex registered at birth.

Approaches 1 and 2: exploratory research

The starting point for the exploratory research was to develop our understanding of the respondent interpretation of the 2011 Census question (PDF 1.74MB) on sex. This was investigated in conjunction with the Equality and Human Rights Commission (EHRC) guidelines for collecting information on gender identity, published in May 2012.

We began our testing with qualitative investigation of three question types using both focus groups and in-depth interviews (2017: 6). The questions types tested were:

• 2011 Census sex question: “What is your sex?” 1. Male, 2. Female
• non-binary sex question: “What is your sex?” 1. Male, 2. Female, 3. Other
• two-question design: “What is your sex?” 1. Male, 2. Female, followed by “Which of the following options best describes how you think of your gender identity?” 1. Male, 2. Female, 3. In another way

The 2011 Census sex question

This work identified the following for the 2011 Census sex question:

• many people answer this question easily, as their sex and their gender identity are the same; however, where they identify their gender as different from their sex registered at birth, the question can be interpreted in different ways (for example, legal sex, sex registered at birth or gender)
• those interpreting the question as sex registered at birth, particularly trans respondents, did not consider the question to be acceptable or relevant; they considered it intrusive
• the question did not identify trans and cisgender people separately
• it was not possible for respondents to record that they are non-binary or intersex

Some participants noted that the display of the response options, with “Male” first, did not conform to rules applied to other questions. Response options are displayed alphabetically, or with the largest group first, unless there is a reason to order response options in a different way. It was therefore suggested that the response options be reordered with “Female” first.

Non-binary sex question and two-question design

Interpretation and acceptability

For the non-binary sex question, the “Other” response increased confusion for trans participants around the question’s meaning.

For the two-question design, trans participants recognised the distinction between the concepts of sex and gender identity but considered deriving their trans status by cross-tabulating responses to the two questions to be “underhand”. Participants questioned the need to ask both questions. For some participants, this was because they thought the questions asked the same thing, and for others this was because they did not see a separate data need.

In both designs, it was recommended that the “Other” or “In another way” response options include a write-in option. Without it, the response option was considered to homogenise the trans population and differentiate them from the rest of society. A write-in option would allow all respondents to provide an accurate and specific answer.

Extent of alignment with data need

In terms of the need for gender identity data, both the non-binary and two-question design approaches were improvements to the 2011 Census question design, as some trans respondents could be identified. However, neither would identify all trans respondents.

For the non-binary sex question, trans respondents who answered “Other” would be identified. Those who identified as male or female would not be.

For the two-question design, trans respondents would be identified if they answered the questions differently. However, testing found that trans participants reported their gender identity in both the sex and the gender questions. Reasons for doing this included concerns about their trans status being potentially visible in the census data and various interpretations of what the questions are asking, such as gender identity, legal sex and sex registered at birth.

Impact of adding a write-in option

We added write-in options to the non-binary sex question and two question designs, as recommended from qualitative testing. These were quantitatively tested (2017: 10). The aim of this test was to investigate:

• the impact on overall census response
• the impact on the quality of data collected in the sex question
• the use of the write-in options

The test showed that there was no significant difference in overall response rates between the three alternatives. These ranged from 36% to 40%. The data also showed that having a gender identity question did not affect the quality of the data collected on sex. Incidences of double-ticking, invalidated questions and tampering with questions was very low across all question designs.

However, we noted that the level of non-response in the gender identity question was significantly higher than that observed for the sex question, irrespective of how the sex question was presented.

Recommendations from Approaches 1 and 2

In our December 2017 topic update, we concluded from this work that none of the question designs covered by Approaches 1 and 2, as currently presented, would meet the requirement for a reliable estimate of the trans population.

While the two-question design was assessed as most able to meet user needs, requirements for further development were identified, including:

• guidance on how to answer the first question on the respondent’s sex needed to be clearer
• the second question on their gender identity needed to be substantively different to the first question
• all trans respondents needed to be able to provide their specific gender identity
• the question needed to directly identify the trans population, not allow its derivation by cross-tabulation

Research also showed that the response options “Male” and “Female” should be listed in alphabetical order, with “Female” first.

Approach 3: development – a gender question and a gender identity question

To meet the recommendations from the exploratory rounds of testing, we explored asking a gender question at the first stage, with an “Other” response option and associated write-in (2017:15). This would be followed by a question on whether the respondent was trans. One version specifically included the word “gender” and the other did not. The two question designs tested for the first stage were:

What is your gender?
[ ] Female
[ ] Male
[ ] Other, write in
(Write in gender)

Are you…?
[ ] Female
[ ] Male
[ ] Or do you describe yourself in another way, write in
(Write in gender)

The “What is your gender?” question was well understood by most participants. Some cisgender participants interpreted it as asking sex; however, this did not lead to inaccurate answers as their sex and gender were the same. For trans participants, the question provided the opportunity to provide a specific and accurate answer.

The unspecified “Are you…?” question was not well understood by trans participants. The interpretation issues found in previously tested question designs were made worse. There was general agreement across participant subgroups that the unspecific “Are you…?” question design was too informal for use in an exercise such as the census. This design posed issues for respondents using assistive technologies, such as screen readers, as the question is not a full sentence.

We found that acceptability increased with the inclusion of the write-in response option, as this allows people to write how they personally identify.

No respondents reacted negatively to the ordering of the response options with “Female” first. Most respondents did not notice the change. When it was probed with respondents, they were either neutral or positive about the change.

Recommendations from Approaches 1 to 3

The results from testing this third approach, along with evidence collected in earlier testing, resulted in four recommendations, which were announced in the White Paper Help Shape our Future: The 2021 Census of Population and Housing in England and Wales. Two recommendations related to the topic of gender identity and two related to the topic of sex.

Recommendation 1: Providing information on being trans should be voluntary

In every qualitative test, participants who identified as trans stated that it was important not to force individuals to disclose that they are trans, as it may not be safe to do so. This led to the recommendation that respondents should have the option to not provide an answer.

Two methods of doing this were identified: Parliament passing primary legislation to amend the Census Act 1920 and remove the penalties for not responding for the question on gender identity or adding a “Prefer not to say” response option.

Subsequent testing included a “Prefer not to say” response option. This was later replaced by a “This question is voluntary” instruction, in line with the religion question, which is underpinned in legislation.

Recommendation 2: The gender identity question should only be asked of those aged 16 years and over

The Census 2021 topic consultation (PDF, 729KB) and gender identity workshop showed a data need exists for the gender of people aged 15 years and under. However, participants in our research (2018:12) were concerned about parents providing a response on behalf of other household members aged 15 years or under. Some parents noted that information provided by parents about children could not be retracted if the child later decided they did not want to be recorded in trans data.

This was explored in research (2017:1) into the public acceptability of asking about gender identity. This research used a recommended gender identity question from the EHRC. It showed that the proportion of respondents stating that asking a voluntary gender identity question was acceptable or very acceptable was lower when considering answering on behalf of someone aged 15 years or under. In particular, considering answering on their own behalf, 80% in England and 75% in Wales said asking a voluntary gender identity question was acceptable, while considering answering for others in the household aged 15 years or under, 69% in both England and Wales said asking a voluntary gender identity question was acceptable.

In May 2018, we held a stakeholder meeting to test the assumption that the minimum age at which the gender identity question should be asked on Census 2021 is 16 years old. Although the stakeholders reiterated the user need for data on gender identity at lower age groups, there was general support for asking the question to those aged 16 years and over only.

The main justification for asking the question to only those aged 16 years and over related to data quality. Responses for children aged under 16 years are likely to be completed by the householder, which in most cases is a parent. The householder may not know if the child’s gender identity is the same as the sex registered at birth, which would reduce the quality of the data collected. However, this also highlighted a possible under-reporting of those who identify their gender as different to their sex registered at birth for those aged 16 to 18 years because, as with children aged under 16 years, the householder will generally complete the questionnaire on their behalf.

These respondents will be able to request an individual access code or paper form if they wish to respond separately to the rest of their household. Answers provided on an individual form will replace any answers submitted on the household form.

Recommendation 3: The 2011 Census sex question wording and response options should not be changed

We recommended that the 2011 sex question remains unchanged, apart from placing the “Female” response option first, as “female” is first alphabetically and the largest group in terms of population size.

In 2011, the response options were placed next to each other on the paper form, rather than in a column. This is not possible for the electronic questionnaire, as it reduces accessibility.

As in the 2011 Census, some trans respondents may find the question difficult to answer, so we continue to work on ensuring any guidance is appropriate. Research into the wording for the 2019 Rehearsal guidance is published in the report Guidance for questions on sex, gender identity and sexual orientation for the 2019 Census Rehearsal for the 2021 Census. We are continuing work on further developing the guidance used for the 2019 Rehearsal.

Answering the sex question will continue to be mandatory for all respondents.

Recommendation 4: The sex question should include a guidance note stating that a question on gender identity will follow

A guidance note will be added to the sex question stating that a question on gender identity will follow. This aims to inform those who identify their gender as different to their sex registered at birth that they will be able to identify their gender in a later question. The guidance also helps them to answer the sex question.

In qualitative testing (2018:6), cisgender participants did not have any problems with the inclusion of the guidance note. They either did not read it or understood why it was included without it impacting on their response.

Initially, the guidance note was placed after the response options. The testing showed that respondents did not consistently notice the guidance note in this position. It was therefore moved to between the question stem and the response options. This is the point at which respondents query the question’s meaning, so this position helps the answer process. The terminology “will follow” was added to the qualifications questions and was used again here for consistency of approach across the census form.

In the 2019 Rehearsal (2019:15), respondents to the online form interpreted the instruction as meaning that the next question would be on gender identity. To clarify this, the instruction online has been expanded for Census 2021.

Final recommended sex question for the Census 2021 electronic questionnaire

What is your sex?
A question on gender identity will follow
[ ] Female
[ ] Male

On paper, as all respondents will see the instruction, it reads “A question on gender identity will follow if you are aged 16 or over”. No change has been made to this following rehearsal as respondents can see the next question.

In other quantitative testing (2018:24), we found that the time taken to complete the version of the sex question with the guidance note was only slightly greater than the time taken to complete the sex question without the guidance note. We found that its inclusion significantly decreased non-response to the sex question, from 0.5% of respondents to fewer than 0.2%.

In the White Paper Help Shape our Future: The 2021 Census of Population and Housing in England and Wales, we stated that we would continue to refine a gender identity question design that meets user and respondent needs.

The conclusions of Approach 3 led us to Approach 4. We would retain the 2011 Census sex question, with a guidance note before the response options stating that a question on gender will follow. “Female” will be listed before “Male”. This would be followed later in the form by a question that will identify the trans population and allow respondents to report the gender they identify with.

The second question was developed to avoid reference to the terms “Female” and “Male” or any similarly interpreted terms, such as “Woman” or “Girl”. This was to make the question both direct and distinct from the sex question.

The initial findings from the research and testing on Approach 4 were discussed in the December 2018 census topic update. Within this section, we provide further details of the testing on this approach and how this led to a viable gender identity question recommendation for Census 2021.

Approach 4: development – a sex question with guidance that a gender identity question will follow and a gender identity question

Community testing

Three distinct core designs were developed and evaluated via community testing at LGBT History Month events (2018:3). These events allowed us to reach a wide range of respondents in a safe environment and in a short period of time. This included participants whose gender was not the same as the sex registered at birth and cisgender participants. These core designs were:

• “What do you consider your gender to be?” 1. Same as sex registered at birth, 2. Different from sex registered at birth, write in gender, 3. Prefer not to say
• “Is your gender the same as the sex you were registered at birth?” 1. Yes, 2. No, write in gender, 3. Prefer not to say
• “Do you consider yourself to be trans or have a trans history?” 1. Yes, 2. No, write in gender, 3. Prefer not to say

The community testing included a number of modifications to these three core versions, such as:

• including a write-in box for those with a gender other than “female” or “male”, compared to not providing an opportunity to specify a gender
• stating that the write-in box aims to collect gender, compared to leaving this open to respondent interpretation
• including a definition of the word “trans”, where relevant, compared to leaving this open to respondent interpretation
• asking about gender and sex registered at birth being different, compared to asking about them being the same

During these community testing events, participants were given a short interview to discuss their views on a random selection of two or three of the proposed questions. The modifications to the questions were added and removed over the course of these events in line with research findings. In parallel, variants of the core designs were cognitively tested (2018:6) with a smaller number of participants.

Results for core versions

The question format “What do you consider your gender to be?” was the least well understood. It included two concepts (gender and sex), with the second being introduced in the response options. Respondents found this confusing because they already had their answer in mind when they read the response options. A small-scale quantitative test (2018:12) validated this finding, with respondents taking, on average, 14 seconds to answer this version, compared to 5 seconds for both alternative versions.

The question format “Do you consider yourself to be trans or have a trans history” was also not well understood. The phrase “trans history” was intended to mean people who have fully transitioned to a binary gender and might not still identify as trans. However, some respondents interpreted it as meaning that they had previously had a trans identity and had transitioned back to the sex registered at birth. In addition, non-binary individuals did not consistently include themselves, even when a definition was added.

In both of these designs, the word “consider” was strongly disliked.

The question most consistently understood was “Is your gender the same as the sex you were registered at birth?” Respondents answered correctly, although some still found the inclusion of sex and gender within a single question to be confusing.

The term “assigned” was suggested as an alternative to the word “registered”, as this term is more commonly used among the trans community. However, “registered” was understood as relating to the act of registering a birth with the sex recorded on the birth certificate. The term “assigned”, in contrast, was interpreted as the sex recorded on the birth certificate or as the sex recorded on medical records, which could include “indeterminate”.

Results for specific design features included in variations

Including a write-in box

Across all tested question designs, participants agreed that an opportunity to provide an accurate and specific response was necessary for the question to be acceptable. This was especially important when in combination with a first stage question that did not allow for expression of an identity other than female or male.

Stating that the write-in aims to collect gender

The question does not directly ask about gender in the question stem. Therefore, in variants where “write in” was used, rather than “write in gender”, participants were unsure what they were being asked to write.

Adding a definition of “trans”

In variants where “trans” was directly referenced, participants had a varied interpretation of the meaning. Non-binary participants did not identify with the term and excluded themselves. Because the sex question only collected “female” and “male” responses, this population would not be identified if a definition was not provided.

Asking about difference, rather than similarity

“Is your gender different to the sex you were registered at birth?” was tested. Participants expected the “Yes” to be the majority answer and found this variant difficult to answer. Some participants suggested a variation using the term “align”, as this is commonly used in the trans community. However, this term was not well understood by cisgender participants.

Quantitatively testing a separate gender identity question

In the December 2018 census topic update, we stated that two different questions were being considered and quantitively tested to help understand the effect on response rates and data quality, along with public acceptability testing. The earlier testing of Approach 2, conducted through cognitive interviews and community testing at LGBT History month events, had led to the decision to take forward the following questions approaches for quantitative testing: “Do you consider yourself to be trans or have a trans history?” and “Is your gender the same as the sex you were registered at birth?”.

“The Office for National Statistics (ONS) undertook a large-scale post-out test (2018:24) of two different questions based on these approaches. Participants were initially sent an access code for an online form. If they did not complete the online form, they were later sent a paper questionnaire.

There were two treatment groups and a third control group. The control group was not asked a question on the gender identity topic but were asked the 2011 Census sex question. The following decisions were applied to both treatment questions:

• question made voluntary via a “Prefer not to say” response option
• question only asked of those aged 16 years and over
• both to use the term “registered at birth”, one in a definition of trans and the other in the question wording
• state that the write-in is asking for gender

Both treatment groups also included the 2011 Census sex question with the addition of the guidance note that a question on gender will follow.

Treatment A

For this question, we removed the term “trans history” and added a definition of the term “trans”. The question included in this test was:

Do you consider yourself to be trans?
Here trans means your gender is different from the sex you were registered at birth
[ ] No
[ ] Yes, write in gender
(write-in)
[ ] Prefer not to say

Treatment B

For this question, we used the term “the same as” to ask about similarity. Alternative terms used by the trans community, such as “aligned with”, were not well understood. The question included in this test was:

Is your gender the same as the sex you were registered at birth?
[ ] Yes
[ ] No, please write in gender
(write-in)
[ ] Prefer not to say

Quantitative test results

The test (2018:24) response rate was higher than we would expect for a voluntary, self-completion, mixed-mode (paper and online) survey with no field follow-up. This gives weight to the robustness of the results. The test had four main aims, and the findings were as follows.

Impact on overall response

There was found to be no significant difference between the three treatment types (Control, Treatment A and Treatment B) in terms of overall response to the questionnaire, although Treatment A had a lower response than Treatment B. The Control and both treatment groups had an overall response rate of between 33.7% and 34.9%. There was also no difference in the proportion of returns online and on paper across the three groups.

Only 0.1% of participants in each of the sample groups stopped completing the survey altogether at the gender identity question.

Impact on response to the sex question

There was a significant decrease in item non-response to the sex question for the groups who were asked an additional question on gender identity compared with the group who were only asked the sex question. In other words, including the gender identity question led to more people answering the sex question than when it was not included. In the Control group, 0.47% of participants did not answer the sex question compared with 0.09% in Treatment A and 0.18% in Treatment B. This included those who selected “Prefer not to say” and those who did not provide an answer at all.

Response distributions for the gender identity questions

There was a significant difference between the treatment types in item non-response to the gender identity question. In Treatment A, 4.74% did not provide an answer compared with just 0.83% in Treatment B. The only other question in the test where a “Prefer not to say” response option was included is the question on sexual orientation.

In both treatment groups (2018:24), the proportion answering that they were trans was 0.1%. The intention here was to show whether the question could identify a trans population, which this achieved. The sample used was not designed to produce a reliable estimate of the trans population.

Public acceptability of the gender identity question

There was a significant difference in acceptability of the gender identity questions between the treatment types: 71% of participants found the Treatment A question acceptable or very acceptable, whereas 90% of participants found the Treatment B question acceptable or very acceptable.

We concluded that the Treatment A question “Do you consider yourself to be trans?” was not suitable for use in England and Wales because of low public acceptability.

We therefore used the Treatment B question “Is your gender the same as the sex you were registered at birth?” in the 2019 Rehearsal.

In parallel, the National Records of Scotland (NRS), which are responsible for conducting the census in Scotland, tested a range of questions aligned with Treatment A, designed to meet a clearly identified need for information on the trans population in Scotland. Their testing showed this approach was acceptable to members of the trans community and the general public in Scotland and produced good quality data. Hence this approach was taken forward in Scotland. Information on this testing is available in the NRS Sex and Gender Identity Topic Report. Annex 4 provides an overview of the differences between the NRS and ONS question designs.

Making the gender identity question voluntary

Three methods of making the gender identity question voluntary were identified:

• Parliament passing primary legislation to amend the Census Act 1920 and remove the penalties for not responding to the question collecting gender identity; the instruction “This question is voluntary” would be included on the question (this is the same approach used for the voluntary “religion” question)
• adding a “Prefer not to say” response option
• implementing both methods: Parliament passing primary legislation, adding an instruction stating, “This question is voluntary”, and adding a “Prefer not to say” response option

Initially, we proceeded with testing based on the non-legislative option of adding a “Prefer not to say” response option. We later investigated the legislative options. A small-scale online quantitative test was conducted to examine the impact of including both a “Prefer not to say” response option and an instruction statement (“This question is voluntary”) compared with just the instruction statement.

The quantitative test (2019:4) found that question non-response was 3.8% with a “Prefer not to say” option and a “This question is voluntary” instruction (this percentage includes “Prefer not to say” responses) and 2.4% with only a “This question is voluntary” instruction.

To collect the best quality data, the non-response should be minimised. We therefore recommended making the question voluntary by Parliament passing primary legislation to amend the Census Act 1920 and remove the penalties for not responding to a gender identity question. This amendment was passed by Parliament in October 2019. We have added an instruction stating “This question is voluntary” with the word “voluntary” presented in bold font.

Question design for online

In addition to the testing described earlier, both the sex and gender identity questions have undergone significant user experience (UX) testing (2017:2, 2018:2, 2019:1 and 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 testing has taken place on a rolling basis since 2017. Before the Census Rehearsal in October 2019, 458 interviews were conducted at 99 events. The UX testing programme will continue through 2020. All participants are purposively selected to include a wide range of ages and digital abilities.

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

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 Welsh Government. The group 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.

None of the participants in any of the Welsh language qualitative tests identified as being trans.

In this testing, the words “rhyw” (sex) and “rhywedd” (gender) were considered too similar. Several participants did not know the word “rhywedd”. Suggestions were made to just use “rhyw”, but this was assessed as changing the meaning of the question.

Given the similarity of the two words in the Welsh language, we considered including an instruction defining gender on the Welsh language version of the form. This would be placed between the question stem and the response options. However, we concluded that the confusion did not impact on participants answering the question correctly. We expect that those whose gender is not the same as the sex registered at birth, who choose to answer in Welsh, would be aware of the two terms.

The gender question was described as “unexpected” by participants. This may be because of its position within the test questionnaire. It appeared at the start, immediately after address and sex, but before age and other demographic characteristics. This should be addressed by its location after the sociocultural topics in Census 2021.

The question designs recommended for Census 2021 have been informed by the research and testing detailed in this report. As is normal practice, the Office for National Statistics (ONS) has finalised the wording of the census questions that were included in the census regulations.

The gender identity question will read “Is the gender you identify with the same as your sex registered at birth?” The questions and response options for Census 2021 have now been finalised through the census secondary legislation. Details of the guidance used in the 2019 Rehearsal, and a summary of the research that informed its development, was published in September 2019 in Guidance for questions on sex, gender identity and sexual orientation for the 2019 Census Rehearsal for the 2021 Census. This guidance is being reviewed following the 2019 Rehearsal, and guidance text or instructions may change if there is enough evidence to support doing so.

Sex

Sex data will continue to be collected in a way that is consistent with previous censuses.

Figure 1: Online question on sex, as asked if aged 16 years or over, in English and Welsh

Source: Office for National Statistics – Sex and gender identity question development for Census 2021

Download this image Figure 1: Online question on sex, as asked if aged 16 years or over, in English and Welsh

.png (71.7 kB)

Design features

• The wording of the question stem and response options is unchanged from the 2011 Census.
• The response options are ordered as “Female” then “Male”.
• The question includes an instruction that a question on gender identity will follow to increase the acceptability of this question; on the electronic questionnaire, this instruction will only appear for respondents aged 16 years and over, as respondents aged under 16 years will not be asked the gender identity question.
• When a respondent is replying on behalf of someone else on the electronic questionnaire, their name, rather than “your”, is included in the question.

Gender identity

The gender identity question enables people to identify that their gender is different to the sex registered at birth, if they want to. This includes, but is not restricted to, those with the protected characteristic of “gender reassignment”.

Figure 2: Online question on gender identity in English and Welsh

Source: Office for National Statistics – Sex and gender identity question development for Census 2021

Download this image Figure 2: Online question on gender identity in English and Welsh

.png (125.8 kB)

Design features

• The question on the gender identity topic is separate to the question on sex and substantively different in phrasing of both question stem and response options.
• Providing information on gender identity is voluntary (an instruction on the question reads “This question is voluntary” and other guidance materials emphasise this); the Census (Return Particulars and Removal of Penalties) Act 2019 removes any penalty for not responding to a question or questions on gender identity in the census.
• The question is only asked to respondents aged 16 years and over; on the online questionnaire, respondents aged 15 years and under are not presented with this question, while on the paper questionnaire, they are directed around this question by instructional text.
• The question provides an opportunity for respondents to state their gender identity; if a respondent replies that their gender is different to their sex registered at birth, they can use the text box to provide their specific gender.
• When a respondent is replying on behalf of someone else on the electronic questionnaire, their name, rather than “your”, is included in the question.

Terminology

The term “gender identity” will continue to be used to describe the topic this question relates to, as this was found to have high acceptability and was easier to understand by respondents than the term “trans”.

The write-in response enables people, if they wish, to identify their gender where this is different from the sex registered at birth. The term “registered at birth” is used to ensure reference to what was put on the birth certificate at that point in time.

Individual forms

Respondents aged 16 years and over are able to request an individual electronic or paper questionnaire if they wish to respond separately to the rest of their household. This enables people to answer the census privately, without having to tell the person completing the household form they have done so. Individual answers will override any answers submitted on the household form. This is vital to protect people’s privacy and ensure good quality data.

Outputs

The question on sex will allow the production of outputs that are comparable across the UK, with other data sources and with previous censuses.

The gender identity question will allow the measurement of the size of the trans population for equality monitoring for the first time. While the question is different to that being proposed in Scotland, these questions were designed to meet the user need identified in that country and to allow the delivery of harmonised UK-level outputs.

This question will also enable the collection of information on the gender of those whose gender identity is different from the sex they were registered at birth. This meets the respondent need to identify how they wish.

Ordering the questions

The sex question is located at the start of the household form where individual data are collected. The new guidance note, stating that a question about gender identity will follow later in the questionnaire, is only relevant to those aged 16 years and over. To allow the guidance note to appear only for relevant respondents, the order of the age and sex questions has been switched so that date of birth is collected before sex.

Census 2021 will be the first to ask questions on both gender identity and sexual orientation, subject to Parliamentary approval, so their position within the form needed to be decided.

Qualitative testing has consistently shown that trans respondents prefer the sex and gender identity questions to be together. However, there is a risk that respondents stop answering the form at the gender identity question. Evidence from quantitative testing showed that only a very small proportion of respondents stopped completing partway through the form. The gender identity question had similar rates of respondent drop-off as other questions.

Given that the position of the gender identity question did not appear to affect respondent drop-off, we were guided by the practical consideration of where the question would best fit on paper. To reduce the complexity of the paper form, we decided that the gender identity and sexual orientation questions should be placed with the questions that only apply to people aged 16 years or over. Therefore, we have placed this question at the end of the sociocultural questions.

The gender identity question is placed with the sexual orientation question, as these two topics are often discussed together. It comes before the qualifications section. This means that all questions for those aged 16 years or over are grouped together.

We carried out a small-scale quantitative test (2019:3) to assess whether the data quality will be improved if the gender identity question or the sexual orientation question is asked first. The test showed that question non-response across the two questions was smaller when sexual orientation was first, compared with when gender identity was first. However, the question that was asked first had higher question non-response than the second. We therefore recommended that the topics should be ordered with sexual orientation preceding gender identity.

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.

Table 1 provides the original assessment scores from May 2016 for the topics of sex and gender identity. The initial assessment of the sex question as “Low” is consistent with the conclusion that the question would capture reliable information. The initial assessment for gender identity was prior to the recommendation for inclusion.

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 summarises the assessment scores for the topics of sex and gender identity from this updated evaluation. We present the evidence used to assess questions as having a “Medium” or “High” potential for impact on the evaluation criteria in Annex 3.

Across all aspects of the evaluation, the assessment has either remained unchanged, where it was assessed as “Low”, or improved. Our position is that the questions being proposed for Census 2021 will capture reliable information.

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), which are responsible for conducting the censuses in Scotland and Northern Ireland, respectively.

We provide an overview of the question development process for NRS in Annex 5. Census 2021 in Northern Ireland will not include a question on gender identity.

We recognise that each country 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.

There is work underway across the Government Statistical Service (GSS) to harmonise measures of sex and gender in data collection across government. Current projects are outlined on the sex and gender page of the GSS website.

The gender identity questions were developed for use in the context of Census 2021 in England and Wales, a mandatory household form. Therefore, it is possible that in different contexts, such as social surveys, a different approach may be more suitable.

We will continue to publish work on the development of a question on this topic on the Gender identity pages of the ONS website.

Sex

In the Census 2021 topic consultation response, the Office for National Statistics (ONS) made clear commitments to the public. We committed to continuing to ask the question on sex. We made two further commitments related to the question on sex.

Review question guidance

The guidance has been reviewed by a broad range of stakeholders and tested with members of the public. The research and the 2019 Rehearsal guidance have been published in the report Guidance for questions on sex, gender identity and sexual orientation for the 2019 Census Rehearsal for the 2021 Census. This guidance is being reviewed further following the 2019 Rehearsal.

Consider improvements possible via online collection, to improve data quality or reduce the burden placed on respondents

Testing suggested two improvements to the sex question as discussed in this report. First, a guidance note was added for those aged 16 years and over making it clear that a question on gender identity will follow later in the form. Online functionality will be used to ensure this is not presented to respondents aged 15 years or under, to whom it is not relevant. Secondly, the response options will be ordered “Female” then “Male”.

Gender identity

As the ONS had identified a clear user need for data on gender identity, we included a gender identity research and testing plan (PDF, 799KB) in the consultation response. This proposed a series of next steps to begin our research into whether to include a question on gender identity in Census 2021. These are listed in the following subsections, alongside an update on our progress towards meeting these commitments.

Identify and plan work to produce gender identity estimates

The research plan was developed, reviewed and updated after each phase of testing. This steered the work outlined in this question development report.

Review the Trans Data Position Paper

The results of this review were published in the report Gender identity update in January 2017.

Seek to learn from other national statistics agencies

The ONS has worked closely with other National Statistical Institutes (NSIs) in developing questions on gender identity. In February 2019, the international position was published as a collaborative UN Economic Commission for Europe paper between Canada and the UK titled, In-depth review of measuring gender identity. This paper provides information on many countries that have either begun measuring gender identity or are looking to do so.

Engage with relevant stakeholders, including members of the trans and non-trans community, to clarify data requirements

This work started with a gender identity workshop in August 2016. We have continued to identify and engage with stakeholders, both formally through events such as gender identity update events in June 2017 and informally through correspondence and individual meetings.

In February 2019, we published What is the difference between sex and gender?, in relation to our progress against the UN Sustainable Development Goals.

Identify alternative options for meeting the user requirement for data

In the December 2018 census topic update, we confirmed that there are currently no administrative sources that record transgender identities, including non-binary, for the whole population. Therefore, we cannot meet the user need through administrative data.

A report, Exploring existing data for gender identity and sexual orientation, details our work on administrative data sources in relation to gender identity. This report will be published on the ONS website concurrently with this question development report.

We are exploring options to include a gender identity question on other surveys. A question was included on the Crime Survey in October 2019.

Development and implementation of options evaluation criteria

Evaluation of research and testing results led to the recommendation that a question on gender identity should be included in Census 2021. Public acceptability was one of the main criteria, and the December 2018 census topic update provided the high-level findings from public acceptability research on the most recent iterations of the question.

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 the sex question

For the sex question, the potential for impact on data quality, public acceptability, respondent burden and financial concerns have been assessed as “Low”. This is the same outcome as in the May 2016 evaluation (PDF, 796KB).

Question development work has improved data quality and public acceptability by making it clear to respondents that a question on gender identity will follow later in the form and by ordering the response options to show parity between female and male.

Potential for impact on questionnaire mode: “Medium”

This question asks for sensitive data from the respondent. The electronic questionnaire has radio buttons for each response option, and on the paper questionnaire these are tick-boxes.

On paper, the instructions for this question are “A question about gender identity will follow if you are aged 16 or over”; however, online, the instruction “A question about gender identity will follow later on the questionnaire” will only appear if the respondent is aged 16 years or over.

Evaluation of the gender identity question

The potential for impact created by the inclusion of the gender identity question is lower for all criteria than in May 2016, except questionnaire mode.

Potential for impact on data quality: “Medium”

This question has not been asked on a census before, and respondents have the option to write in a response. This question asks sensitive information that can be difficult to answer on behalf of another person.

Feedback from the 2019 Rehearsal (2019:15) showed that some respondents found this question difficult to answer, but the non-response rate was lower than expected.

Potential for impact on public acceptability: “Medium”

The responses to this question could be at risk of social desirability bias, when a response is given that is considered socially desirable rather than accurate. This question asks for information that a respondent may not want answered by a proxy on their behalf.

Testing (2018:24) showed that 90% of respondents found a gender identity question on Census 2021 acceptable. Only 0.1% of respondents stopped completing the survey altogether at the gender identity question. Qualitative testing indicated that some of those who did not find it acceptable did so as they consider gender and sex to be the same concept. Some respondents to the 2019 Rehearsal (2019:15) also indicated that they found the question unacceptable.

To reduce the potential impact on public acceptability, the gender identity question is both voluntary and only asked of people aged 16 years and over. We have also included a write-in option to allow all respondents to report their gender identity. Respondents will be able to request an individual access code or paper form if they wish to respond separately to the rest of their household.

Potential for impact on respondent burden: “Medium”

This question has a write-in response, and this question asks for information that cannot be observed and so can be difficult to answer on behalf of another person. Some respondents to the 2019 Rehearsal (2019:15) reported that they found this question difficult to answer.

Potential for impact on questionnaire mode: “Medium”

This question asks for sensitive information. This means that there could be a risk from impact on questionnaire mode, as respondents may be more reluctant to give an answer on the paper questionnaire, which could be seen by other members of their household, than on the electronic questionnaire, where answers are more difficult to find. However, all respondents can request an individual paper questionnaire or request access to an individual online questionnaire, so the risk has been mitigated.

Sex and gender are terms that are often used interchangeably, but they are in fact two different concepts.

The sex question is binary: female and male.

The gender identity question is about a person’s personal internal perception of themselves. As such, the gender category with which a person identifies may not match the sex they were registered at birth.

Throughout this report, we have used the widely used word “trans” to describe all those whose gender is not the same as the sex they were registered at birth. Language around this topic is still emerging. Our use of the term is inclusive of a range of genders, including:

• binary male or female genders when not the same as registered at birth
• non-binary genders such as those on a continuum between male and female
• non-gendered identities (neither male nor female)

These genders may be fixed or variable.

The term “cisgender” is used to describe those whose gender is aligned with the sex they were registered at birth. Cisgender includes male when registered male at birth and female when registered female at birth.

Variations of sex characteristics, sometimes known as intersex, is not a gender identity, but it can lead to a person having a gender that is not the same as the sex they were registered at birth.

Throughout the development of the Census 2021 questionnaires, the Office for National Statistics (ONS), National Records of Scotland (NRS), and Northern Ireland Statistics and Research Agency (NISRA) have worked together to ensure harmonisation of the outputs from the sex and gender identity questions, where possible. You can learn more about this ongoing harmonisation work on the Government Statistical Service (GSS) sex and gender web page.

Under the current proposals from each of the statistics agencies, the data gathered on the important demographic variable “sex” will be harmonised across the UK. The ONS has further developed a census question to measure gender identity. NRS have developed a question to measure trans status or history.

In Northern Ireland, Census 2021 will only include a question on sex, it will not include a question on gender identity.

Gender identity question versus trans status question

The ONS has recommended that a gender identity question is used in Census 2021 in England and Wales. The testing and research that led to this decision is documented in this report. The NRS have proposed a trans status or history question for Scotland’s Census 2021. The NRS made this recommendation after consultation, stakeholder engagement, and research and testing of potential question designs.

Stakeholder engagement by the NRS identified a clear user need for data on the trans population in Scotland. Stakeholders’ preferred method was asking a direct question if a person is trans, rather than the alternative of asking a question about sex assigned at birth followed by a gender identity question.

The NRS qualitatively and quantitively tested a range of potential gender identity questions to understand acceptability and data quality. Their testing found that the trans status or history question was acceptable to members of the trans community and the general public in Scotland and that this question produced good quality data.

For further details, see the NRS’s Sex and Gender Identity Topic Report published in September 2018. This report provides further details of the research and testing conducted by the NRS to help shape their proposed question to collect data about the trans population. In December 2019, the NRS published a Sex Question Recommendation Report, along with a report by ScotCen on testing the guidance of the sex question. After considering all of the evidence relating to the sex question and accompanying guidance, the NRS continue to propose a binary sex question with self-identification guidance for Scotland’s Census 2021.

Ordering of questions

The ONS has recommended that the gender identity question is placed at the end of the sociocultural questions, rather than immediately after the sex question.

The NRS have proposed that the trans status question will be placed on Scotland’s census questionnaire immediately after the sex question. This decision was made by the NRS as testing showed that presenting the sex and trans status questions together provided better understanding by respondents and improved data quality.