Scotland’s Census 2022: Lessons learnt for the census questions and questionnaires

Introduction/Background

National Records of Scotland (NRS) undertook a lot of research and testing to make sure Scotland’s Census 2022 asked questions that met identified user needs. We ran a programme of activity to engage with people and organisations across Scotland to help us decide what subjects to cover and what questions to ask in the census. In developing questions, we considered:

• how acceptable questions are to the public
• how to ask questions in a way that produces reliable answers
• whether other ways of collecting the information already exist. 

More information on how we developed the census questions for 2022 is available on our website. We ran a topic consultation in winter 2015-16. Our topic reports explore the results of our research into individual topics. They cover the background and evidence that went into designing questions for the 2022 Census.

We took care when designing both the online and paper census questionnaires. As expected, most people (89% of those households that returned a form) completed the census online in 2022. We provided a paper questionnaire to households who could not, or did not wish to, complete the online questionnaire. We designed the online and paper questionnaires to be accessible and as easy to use as possible. This was backed by research, testing, and best practice. People could ask for translated guidance, to complete on the phone through our contact centre or complete online in Gaelic if they wanted to. Laterally, doorstep capture was also made available.

You can view and download the complete question set from the 2022 census for each of the census questionnaires.

Census data was quality assured at every stage of collection, the data processing and outputs publication journey. As a result of this quality assurance, a number of lessons were learnt on the performance of each census question. This document details these lessons and associated quality issues for questions across the three main census questionnaires (individual, household and communal establishment manager).

We have published documentation on how we decided which questions to include and how they should be asked. To do this, we used five evaluation criteria, namely: strength of user need, suitability of alternative sources, acceptability, clarity and data quality, comparability and operational considerations. 

Planning for Scotland’s Census 2031

Planning has begun for the next Census in 2031. On 17 June 2025 we opened our topic consultation to ask users for their views on the 2022 Census outputs, how far they met their needs and about future requirements for census topics.

NRS will use the responses to the consultation along with broader stakeholder engagement to inform our work on deciding which topics to be included in the 2031 Census and when designing questions to best meet these user needs.

A key lesson from Census 2022 is that we should design in the use of administrative data from the outset. In 2022, administrative data was used successfully in a number of ways and was key to producing the high-quality outputs that covered 100% of the population. In 2031, we plan to use administrative data more widely, for example, in planning operations and contact strategies and in enhanced outputs through linkage to other sources.

List of questions with no lessons learnt to record

Questions that do not need reconsidered as part of future Census data collection exercises in Scotland. This does not necessarily mean they will be used again in the next census, just that there were no lessons to record from 2022.  Topics and questions will be determined through user consultation and testing.

Household questionnaire

H1. Who usually lives here? 
H2. Counting everyone you included in question H1, how many people usually live here?
H3. Starting with the householder(s), list the names of the people counted in question H2, including children and babies.
H4. Is there anyone staying at this address on the night of 20 March 2022 whose permanent or family home is elsewhere?
H5. Counting only the people you included in question H4, how many people are staying at this address on the night of 20 March 2022 whose permanent or family home is elsewhere?
H7. What type of accommodation is this?
H8. Are all the rooms in this accommodation behind a door that only this household can use?
H10. What type of central heating does this accommodation have?
H11. [Routing instructions]

Individual questionnaire

I1. What is your name?
I3. What is your sex?
I7. During term-time, do you live: 
at the address on the front of this questionnaire?
at another address?
I8. Which of the following best describes your sexual orientation?
I10. If you were not born in the United Kingdom, when did you most recently arrive to live here?
I15. Can you use British Sign Language (BSL)?
I17. How is your health in general?
I22. What do you feel is your national identity?
I24. [Routing instructions]
I26. Have you previously served in the UK Armed Forces?
I27. In the last seven days, were you doing any of the following?
I28. Which of the following describes what you were doing in the last seven days?
I29. In the last four weeks, were you actively looking for any kind of paid work?
I30. If a job became available now, could you start it within two weeks?
I31. In the last seven days, were you waiting to start a job already accepted?
I33. Answer the remaining questions for your main job or, if not working, your last main job.
I34. In your main job, what is (was) your employment status?
I35. What is (was) the name of the organisation or business you work (worked) for?
I37. Briefly describe what you do (did) in your main job.
I39. Do (did) you supervise or oversee the work of other employees on a day-to-day basis?
I41. [Routing instructions]
I43. How do you usually travel to your main job or course of study (including school)?
I44. [Routing instructions]

Communal establishment manager questionnaire

C1. What is the nature of this establishment?
C2. [Routing instructions]
C5. Who is responsible for the management of this establishment?

Questions with lessons learnt to record

In this section, for each remaining question, we record lessons learnt and lessons to consider for any future Census.

Household questionnaire

Question H6. How are the members of this household related to each other? 

The Census questionnaire includes a relationship matrix allowing each individual to indicate their relationship with all other individuals in the household. The online version of the relationship matrix used in 2022 was a big improvement from 2011. The 2022 version allowed an individual’s name to be visible when completing the matrix, preventing confusion between people in the household and mistakes. 

However, the matrix still becomes increasingly difficult to fill in correctly and consistently as household size increases. This is particularly the case for paper returns. Although there are steps in the data processing journey to correct detected inconsistencies it is not always possible to do so. Some, but not all, household composition variables rely on the relationship matrix and errors in the matrix can lead to inconsistencies among composition variables depending on the input data.

To mitigate some of these issues, for the Online Collection Instrument (OCI), NRS should test adding more validation to ensure inputted relationship information matches later questions. For example, ensuring that people who respond that they live with their spouse have a marital status of married).

It is also recommended that testing is carried out on  harmonising the different relationship terms with the England & Wales 2021 Census, for example using the term ‘legally registered civil partner’ rather than ‘registered civil partner’, and adding in an option to indicate half-brother / sister. It could also be worthwhile testing the amendment of ‘partner’ to ‘cohabiting partner’ to minimise confusion between ‘registered civil partner’ and ‘partner.’ Space constraints on a paper form will also need to be considered.

Question H9. How many bedrooms are available for use only by this household?

For online returns, it is recommended that we test hard validation (individual is unable to proceed to the next question) to prevent a household from having ‘zero’ bedrooms and adding soft validation (individual is asked to confirm their response) if more than 7 or 8 bedrooms are entered. The guidance for 2022 stated respondents should count a bedsit or studio flat as having one bedroom, but not everyone may have accessed the guidance. 

To improve downstream data processing, further quality assurance should be performed on census returns with large numbers of bedrooms. For example, paper capture scanning errors sometimes led to increased numbers, such as ‘5’ being captured as ‘50’. 

Question H12 Does your household own or rent this accommodation? and H13 Who is your landlord?

During quality assurance of the Census 2022 data, we identified that most households who responded that they ‘lives here rent free’ were in the socially rented sector. It is generally not possible to live in accommodation in this sector without rent being paid. It is possible for rent to be paid through benefits directly to the social landlord, which may have caused people to have responded that they live rent free in error due to them not directly paying rent to their landlord. Redesigning these questions or adding validation to the questionnaire should be tested to reduce the occurrence of this error.

In addition, administrative data on households in the socially rented sector was used for quality assurance in 2022. NRS should investigate whether it could be used to supplement the data collected from these questions in the 2031 Census. One of the uses for this would be to remove this type of error from the ‘lives here rent free’ category. 

Question H14. In total, how many cars or vans are owned, or are available for use, by members of this household? 

To save space, this question could be redesigned to match question H9 on number of bedrooms, where a household just writes in a number or there is a tick box for ‘none’ and a write-in box for anything greater than none. This option would be tested before any changes are made.

Individual questionnaire

Question I2. What is your date of birth?

In 2022, a small number of households responded that no individuals present were aged 16 or over. NRS should test adding validation in the online questionnaire if no individual is aged 16 or over in the household.

Question I4. Do you consider yourself to be trans, or have a trans history? 

This was a new question for 2022 and performed well. Having identified the most common responses, NRS should test including tick boxes for ‘no’, ‘trans man’, ‘trans woman’, ‘non-binary’ and ‘other’ (write in) as this would simplify the coding of responses to this question.

Question I5. On 20 March 2022, what is your legal marital or registered civil partnership status? 

More people responded in the 2022 Census that they were currently or formerly in a registered civil partnership than the number of civil partnerships ever to have taken place in Scotland. This number is considered to be implausibly high and a quality note for this data accompanied its publication from the 2022 Census. Some people have likely ticked this option in error when co-habiting with a partner, but not in a legal civil partnership with them. Space permitting, NRS should test adding clarification, possibly through the wording of the civil partnership options on the questionnaire, to ensure that individuals are giving their legal status not their partnership status. 

Question I6. Are you a schoolchild or student in full-time education?

There is an identified user need to be able to distinguish school children from other individuals in full-time education (i.e., those studying at further and higher education establishments). It is noted however that lower user need for level of education or for part-time vs. full-time education was identified in consultation work for the 2022 Census and space constraints and respondent burden will also need to be considered alongside strength of user need from the topic consultation.

Questions I9/I10. What is your country of birth? and If you were not born in the United Kingdom, when did you most recently arrive to live here?

In the online questionnaire, individuals who ticked that their county of birth was Republic of Ireland were incorrectly routed past question 10 on date of arrival. More information on this is available in the 2022 Census Demography and Migration QA report. This should be corrected for future censuses.

Question I11. One year ago, what was your usual address?

NRS should test adding validation to ensure that the response to this question is consistent with question 10 on the date of arrival to live in the UK. For example, in 2022 it was possible for a person to indicate that they had arrived to live in the UK five years ago, but then respond that they lived outside the UK one year ago.

The drop-down list of countries (if outside the UK) available for online respondents functioned well. NRS should test adding a tick box to differentiate addresses in Scotland versus the rest of the UK as the current method of using UK-wide address matching to make this distinction delayed the release of these data from the 2022 census.

Question I12. Do you look after, or give any help or support to family members, friends, neighbours or others?

Our quality assurance processes found evidence to suggest that the 2022 census did not capture carers who carry out a few hours of care each week well.  NRS should test adding an additional category of 1 to 4 hours a week or align categories with the England and Wales 2021 Census where the smallest category is 1 to 9 hours a week.

Question I13. How well can you understand, speak, read and write English?

Given that the ‘very well’ and ‘well’ categories are combined when census outputs are published, NRS should test combining these in the question as well. 

Question I14. Can you understand, speak, read and write Scottish Gaelic or Scots? 

The layout of the paper question led to ambiguity for individuals that had skills in one of Scottish Gaelic or Scots but not in both. Individuals likely left the tick boxes all blank for the language in which they had no skills, since there was no box to indicate no skills in only one of the languages (they could tick a box to indicate no skills in either language).  Space permitting, NRS should test redesigning the question to be two separate questions to avoid confusion.

Question I16. What is your main language? 

Some write-ins indicated confusion on what to supply if an individual could not speak or communicate. NRS should consider providing guidance for such cases.

Question I18. Do you have any of the following, which have lasted, or are expected to last, at least 12 months? 

NRS should test whether it is necessary to have a tick-box for ‘long-term illness, disease or condition’ and ‘other condition’. Many of these were respiratory conditions, which were coded to ‘long-term illness, disease or condition’ in 2022. NRS should test including a tick-box for respiratory conditions only. A large number of the remaining conditions were age-related conditions. It could be worth testing a separate tick-box or including additional guidance on what to do if an individual has an age-related condition. Redesigning this question will require balancing space, coding costs and user needs. 

2022 included new descriptors for the conditions, learning disability, learning difficulty and developmental disorder, but there is evidence to suggest these did not help individuals successfully differentiate which tick box they should tick given their condition. These descriptors should be revised for the 2031 Census following user consultation and testing. 

To provide richer data and act as a validation of the main category response, NRS could test adding a second step where individuals are asked to select their exact condition from a list within the category or categories they have ticked. This approach is only feasible online due to space constraints in the paper questionnaire.

We note that few individuals ticked the new category of full or partial loss of voice or difficulty speaking and we should assess whether this category is needed or whether the space could be devoted to different conditions. 

Question I19. Are your day-to-day activities limited because of a health problem or disability which has lasted, or is expected to last, at least 12 months? 

Some individuals indicated their activities were limited, but had specified in the preceding question that they had ‘no condition.’ In England & Wales 2021 Census, this question was only asked of individuals who indicate they have a long-term condition; NRS could test using the same routing here.

Question I20. What passports do you hold?

This was a new question for 2022. Although guidance was available on how to answer this question if your passport had expired, it may be worth testing including this information on the questionnaire, space constraints allowing, to ensure individuals answer consistently.

Question I21. What religion, religious denomination or body do you belong to?

Every answer to each census question gets a numeric code, which we use to help us produce census results. This is called coding. We use a document called a coding specification that tells us how to code questionnaire responses in a way that we can analyse. NRS should consider updating the coding specification for religion based on write-ins received in 2022 to reduce the need to manually code responses that do not match the specification.

Question I23. What is your ethnic group?

There was evidence that individuals from Italy may have erroneously ticked Roma as their ethnic group. More information on this issue is available in the QA report that accompanied the release of the Census 2022 data on ethnic group. This option should be redesigned or placed elsewhere in future.

The first tick-box from African, Scottish African or British African and Caribbean or Black was removed between 2011 and 2022 due to space constraints. This removal meant that 2011 and 2022 data were only comparable when the two tick-boxes in 2011 were collapsed for each major category. 

NRS will consider stakeholder views and the results of testing on the design of this question, including the ethnic group categories and placement of tickboxes for the 2031 Census.

Question I25. Which of these qualifications do you have?

The categories available and published differed between the 2011, 2022 and England & Wales and Northern Ireland 2021 Censuses. The education systems across the UK and Scotland do differ, but it may be worth testing harmonising options (e.g., level of attainment) with elsewhere in the UK to facilitate comparison.

It may also be useful to test adding a definition of apprenticeship to the questionnaire to aid completion. There is evidence that there was some respondent error in the apprentice options as foundation and graduate had an older age profile than should be possible considering the short period since their introduction.

NRS should also consider testing reordering, removing or relabelling categories capturing ‘other’ qualifications to simplify the question. It may also be worth adding sub-headings to the paper question subject to space considerations (similar to the online questionnaire). 

Question I32. Have you ever done any paid work?

The number of individuals categorised as ‘never worked’ increased substantially from 2011 possibly due to a change in question from ‘have you ever worked?’ to ‘have you ever done any paid work?’ The change was done in response to user need for data on paid work. NRS should consider further user testing to determine how individuals answer this question particularly those that have been retired for a long time or that have worked in the home or voluntarily but never been paid for their work. 

Question I36. What is (was) your full job title (i.e., occupation)?

NRS should consider testing using examples that are options in the drop-down menu options that exist in the coding specification to avoid people writing in something that is hard to code.

Coding for occupation (and industry, see below question 38) was resource-intensive. NRS should consider updating coding approaches, for instance test using AI techniques, and encourage closer interaction between coding and quality assurance processes earlier in the process to ensure successful coding outcomes. In the online questionnaire, consider testing that individuals could be made to choose their closest-match job to avoid manual coding of job titles during data processing. 

Question I38. What is (was) the main activity of your organisation, business or freelance work (i.e., industry)?

Similar to the occupation question above (question 36), coding approaches of industry need careful consideration including testing making individuals  choose from a set of industries to avoid the additional cost of manual coding.

In 2022, industry was coded to SIC 2007 subclass (5 digit level) which is very precise (806 distinct options). Most respondents do not provide enough detail to code their industry at this level. NRS should consider testing whether coding to division (2 digits, 88 distinct options) is adequate for user needs. This approach would reduce the burden on manual coding to classify industries too specifically. 

Question I40. In your main job, how many hours a week do (did) you usually work?

The hour band options do not match the hour bands collected in the main comparator data source, the Annual Population Survey. NRS may want to consider revising to facilitate comparability and to match common definitions of part-time and full-time work. However, any change may affect Census time series analyses so should be balanced with user need for such analyses. 

Question I42. What address do you travel to for your main job or course of study (including school)?

In 2022, there was an issue with higher than expected numbers of schoolchildren who indicated they were distance learning, home schooled or equivalent. This issue was also present in the 2011 Census. It is not clear what the source of this issue is, but could be due to individuals not easily recalling the address of their, or their child’s school and ticking distance learning, home schooled or equivalent to avoid having to look up the address. Possible options to mitigate this issue are user testing to better understand this issue or providing a better address lookup that could auto-populate the address if the name of the workplace or school is inputted.

Communal establishment manager questionnaire 

Question C3/C4. How many female/male residents usually reside at this establishment in each of the age ranges below?

The number of usual residents of a communal establishment should match the total number of individuals provided in the breakdown by age and sex (questions 3 and 4).  NRS should consider including guidance on the form that this should be the case and whether to add a hard validation in the OCI to check that they match.

The age breakdowns on the form are less useful at capturing the age distributions of students and care homes, two of the major types of communal establishment resident. NRS should also consider revising the age breakdowns to better capture communal establishment populations. Another option would be to consider alternative means for CE managers to provide data, e.g., via email or the facility to upload a spreadsheet. 

A number of communal establishment types typically have very few usual residents (e.g., hotels, general hospitals). Although guidance notes are provided to managers on how to fill in the form, it may be worth considering testing putting some guidance directly on the form subject to space to differentiate usual residents from visitors to prevent additional data processing to adjust unlikely usual resident counts.