Census 2021 outputs: content design and release phase proposals

Closes 5 Oct 2021

Section 3: Main changes to variables compared to the 2011 Census

The census has always changed with the times to better meet user needs, through:

  • making better use of technology
  • changing and replacing existing questions
  • introducing new questions where needed 

In the 2015 consultation, The 2021 Census – Initial view on content for England and Wales, we asked users of census statistics for their views on the topics that the Census 2021 questionnaire might cover. The questions asked in the Census 2021 questionnaire reflect the findings of that consultation.

This section describes the changes in which variable classifications will be available, including some cases where there will be slightly less detail. It will also describe cases where we’ve asked new questions and will illustrate the range of detail and insights that will be available to users.

Full details of the work that we carried out to determine the questions for Census 2021 are available on the Question development pages of the ONS website. In most cases, the questions have remained the same. Where there have been changes, these mostly provide either more or the same detail. However, in a few instances, less detail is collected in Census 2021 than was collected in the 2011 Census.

Where the same information is collected, but through a modified question, this is not included in this consultation as it does not impact the outputs. For example, in the 2011 Census there was a single, complex question on qualifications held. For Census 2021, we’ve asked separate questions on degree-level or above qualifications and on apprenticeships. However, this has not impacted the variable classifications. Similarly, we’ve not changed the national identity question and response options, but we’ve changed the order in which they’re presented in England.

Classifications for questions that have not changed will remain largely the same, except where analysis of write-in responses indicates a change is needed.

We also reviewed and redeveloped all question guidance. In some cases, this has slightly impacted the data being collected. For example, there has been a change to the guidance on how to answer the sex question. As these changes do not impact the classifications, we do not discuss them here.

Finally, on the electronic questionnaire, we implemented search-as-you-type and address look-up functionality for many sociocultural questions. We did this to make it easier for respondents to self-define when completing Census 2021 online. Our Search-as-you-type and address look-up functionality for Census 2021 report, published on 21 December 2020, provides more information. Inclusion of a group on the search-as-you-type list does not necessarily mean that outputs will be available for that group.

We expect that societal change since the 2011 Census, and the implementation of the search-as-you-type functionality, will have affected written responses to questions on cultural background. For example, there are likely to be changes in how people listed their main language or national identity. We also added a write-in option for those selecting “African” in the ethnic group question, which will collect new data on this population. We have not included the detailed classifications for these questions in the specification. This is because the written responses in the Census 2021 questionnaire and the user needs raised in this consultation will inform them.

For each of the topics discussed in this section, we’ve detailed the proposed variable classifications in the Draft proposals for outputs data content (XLS, 610KB) spreadsheet. We’re still developing the most detailed classifications based on Census 2021 responses. As a result, this spreadsheet does not include the detailed variable classifications for the questions on sociocultural background. We’ve also detailed the analysis proposals in the Census 2021 analysis programme proposals (PDF, 504.8KB) document.

Questions with detail removed

In a small number of cases changes to the questions asked resulted in less detail being collected. In all cases where the detail had decreased, we assessed that there was insufficient user need to continue to collect this information. This also offset the increased burden put on respondents to provide additional detail, such as on the new topic of armed forces veterans. We summarise some of the important changes under the following seven topic headings.

Number of rooms Questions with detail removed

We removed this question. Instead, we’ve sourced this information from Valuation Office Agency (VOA) data. Reports on our work towards achieving this are available on the housing characteristics page on the ONS website.

Migration intention to stay

We combined the options, differentiating between those intending to stay in the UK for three to six months and six to twelve months, into a single response option.


We collected all the 2011 Census categories but in a restructured series of questions, except for “Professional qualifications”. We asked respondents to choose the nearest equivalent for any qualification not listed.

Unemployment history

We removed the option to write in the year last worked. We replaced it with options that separate those who are not in work but have previously worked into two groups. These two groups are the short-term unemployed, who have worked within the last year, and the long-term unemployed, who most recently worked over a year ago.

Activity last week

We removed the option “On a government sponsored training scheme”. We did this because the International Labour Organisation (ILO) definition of employment has changed since 2011 to exclude unpaid training as a form of economic activity.

Communal establishment resident age group and types

On the communal establishment form, we removed questions on age groups and types of resident catered for. Data on age groups can be derived from the information collected on residents’ individual questionnaires.

Rough sleepers

In 2011, there was an output on “Rough Sleepers” based on a count completed by census field staff. We’ve removed this because of concerns about the quality of the data collected. We discuss options for alternative outputs about the homeless population in “Section 4: Proposals for potential new derived variables”.

New questions and associated classifications

Sexual orientation and gender identity

We asked the questions on sexual orientation and gender identity for the first time in 2021. The questions were voluntary, and we only asked these questions to those who were aged 16 years and older.

We propose to produce variables representing the response options in these questions, subject to the data passing statistical disclosure control assessments once the final data are available. Because of the small size of some of these groups, we’re unlikely to publish the data at Output Area level.

For sexual orientation, the proposed classification is:

  • Straight or heterosexual
  • Gay or lesbian
  • Bisexual
  • Other

For gender identity, the proposed classification is:

  • Yes: Gender identity is the same as sex registered at birth
  • No: Gender identity is different to sex registered at birth

In both cases, there would be a further “not stated” category, which would include any responses that could not be coded and those respondents who did not answer. In line with how we’ve previously treated the voluntary question on religion, we will not impute answers to these questions.

We’ve included the proposed specifications for the ready-made tables, including these topics, in the “Ready made tables” tab of the Draft proposals for outputs data content (XLS, 610KB).These specifications include the geographical levels the data will be available for.  

Given the potential for disclosure being higher for these populations, we need to carefully assess what additional information we can publish and for what geographical levels.

In addition, we’re developing a more detailed classification using the write-in responses from the Census 2021 questionnaire. Because of the small numbers of people likely to be in each category within a more granular classification, we’re considering how best to output this data.

We’d like to know your views on our plans for producing statistics and analysis using responses to the sexual orientation and gender identity questions.

Armed forces veterans

For Census 2021, we asked a new question on previous service in the UK armed forces. We only asked this question to those who were aged 16 years and older. We created a new standard variable to provide a range of outputs on the population who have previously served in the UK armed forces. The proposed classification is:

  • Previously served, Regular
  • Previously served, Reserve
  • Previously served, Regular and Reserve
  • No code required

The “No code required” category includes both those who have never served and those who are currently serving. We’re also investigating the feasibility of two new variables.

The first proposed new derived variable would indicate if the household reference person has previously served in the UK armed forces. This would have the same categories as the standard variable described above.

The second proposed new derived variable would indicate the number of former UK regular and reserve armed forces personnel resident in a household. The simplest proposed classification is:

  • No former UK armed forces personnel in household
  • One or more former UK armed forces personnel in household

However, we’re considering if we should also include a second, more detailed classification. This might, for example, separate out those with one, two, three, four, five, or six or more former UK armed forces personnel in the household. We will consider the upper threshold of this classification based on Census 2021 data, once available, to take into account statistical disclosure risks.

This variable could be cross-tabulated with the number of people in the household to understand the proportion of the household that are former UK armed forces personnel.

We’d like to know your views on our plans for producing statistics and analysis using responses to the question about armed forces veterans.

Proxy answer question

Another new question in 2021 asked all respondents if they were completing the questions for themselves or for someone else. This provides an interesting new analysis variable, allowing researchers to investigate if there are trends in the data created by who completed the form.

This variable will be in the microdata products held within our Secure Research Service (SRS). We’re considering if we should include it in the main data releases.

We’d like to understand if users would use information on if a respondent answered for themselves or for someone else in their analysis.

Other significant changes to output classifications

Ethnic group classifications

In 2011, there were 18 response options in the ethnic group questions. These were divided into five sections. Within each section, there was an “Other” option allowing respondents to express their identity how they wished. This meant that we could further disaggregate ethnic group data based on the written responses in each section.

For Census 2021, the ethnic group question includes a new tick box for “Roma” under the “White” category. It also contains an option to write in a more detailed response when selecting the existing “African” response option.

We will update the ethnic group classifications to take into account these changes. We’ve shared the current proposed classifications in the “Ethnic group” tab of the Draft proposals for outputs data content (XLS, 610KB). As we’re still developing the most detailed classification, this spreadsheet does not include the most detailed ethnic group classification. This will use data from the write-in responses. 

We’re carefully considering how to best present the classifications. For example, we need to decide how to reflect the slight difference in wording for high-level categories between the Wales questionnaire and the England questionnaire. In Wales, the category “Asian or Asian British” is expanded to read “Asian, Asian Welsh or Asian British”. Similarly, the category “Black, Black British, Caribbean or African” is expanded to include “Black Welsh”.

We’re also considering how to order the classifications in the outputs. In the proposed classifications, we’ve ordered the groups as follows:

  • Asian, Asian Welsh, Asian British
  • Black, Black Welsh, Black British, Caribbean, African
  • Mixed or multiple ethnic group
  • White
  • Other ethnic group

For more detailed classifications we propose to list the groups alphabetically within the higher-level categories.

We’d like to know your views on the proposed classifications for the ethnic group variable, including how they’re presented.

We’d also like to know your needs for analysis on this topic, especially around particular subgroups you’re interested in.

Age classifications

In 2011, we used around 55 different age classifications in the standard outputs. The different age classifications aimed to meet users’ needs for different topics. For example, these included employment or education. To reflect our updated approach for dissemination, particularly the new capability for users to build their own tables, we’ve reviewed the classifications needed for each topic.

This work has resulted in a proposal to reduce the number of classifications by almost half. We’ve done this by creating simple groupings of classifications that get progressively more detailed for each topic. We hope this will make navigating the different classifications easier for users.

We’ve detailed these age classifications in the “Resident_Age” tab of Draft proposals for outputs data content (XLS, 610KB).

We’d like to know if this consolidated list of age classifications will meet your needs.

Country of birth classifications

In 2011, we divided the Country of Birth output groupings for EU member countries into two groups according to when those countries joined. These groups were:

  • countries that joined the EU before 2001
  • countries that joined the EU between 2002 and 2011

Since 2011, Croatia has joined the EU and the UK has left the EU. None of the other classifications based on countries, such as national identity or country of second address, contain this grouping. Therefore, we do not plan to reproduce this grouping in Census 2021 outputs. Instead, the three main categories that we propose are:

  • Europe: United Kingdom
  • Europe: Ireland
  • Europe: Other Europe

We will split these into more groups in the detailed classifications.

We’ve included the groups contained within each country classification in the “COB” tab of Draft proposals for outputs data content (XLS, 610KB).

We’d like to know if the proposal to group all EU member countries as “EU member countries” will meet your needs.

Other classifications with more detail added

To reflect changes in questions and response options in Census 2021, we’ve updated further classifications. In general, users can still get equivalents to 2011 data. We provide full details in Draft proposals for outputs data content (XLS, 610KB) but summarise the changes under the following six topic headings. 

Type of accommodation

We’ve added a new category of “a flat, maisonette or apartment: part of another converted building (for example, former school, church or warehouse)”.

Central heating

We’ve split the category “Gas” into “Mains gas” and “Tank or bottled gas”.

We’ve split the category “Solid fuel” into “Wood (for example, logs, waste wood or pellets)” and “Solid fuel (for example, coal)”.

We’ve split the category “Other central heating” into “Renewable energy (for example, solar thermal or heat pumps)”, “District or communal heat network” and “Other”.

Marital Status

A second question followed the marital status question. That second question asked who the marriage or civil partnership was with. This included the options “Someone of the opposite sex” and “Someone of the same sex”.

Second address

We’ve split the category “Other” into “Partner’s address” and “Other”. 


We’ve reworded the questions on physical or mental health conditions or illness to align with the GSS Harmonised standard questions. This means the census now collects data more closely aligned with the definition of disability in the Equality Act 2010.

As part of this process, we split the question into two parts. The first part asked if the respondent had a condition or illness, and the second part asked if it limited their day-to-day activities.

In the 2011 Census, the category “No” encompassed two groups of people. This included those without a condition or illness and those with a condition or illness that did not impact their day-to-day life. Splitting the question into two parts means these two groups can now be identified separately.

Unpaid Care

We increased the range of response options.

We split the option “Yes, 1 – 19 hours a week” into two categories, which are “Yes, 9 hours a week or less” and “Yes, 10 to 19 hours a week”.

We split the option “Yes, 20 to 49 hours a week” into two categories, which are “Yes, 20 to 34 hours a week” and “Yes, 35 to 49 hours a week”.

We’d like to know your plans for using the additional detail allowed by the new response options in Census 2021 questions.