Official Report

Good morning and welcome to the 20th meeting in 2019 of the Culture, Tourism, Europe and External Affairs Committee. I remind members and the public to turn off mobile phones, and any members who are using electronic devices to access committee papers to ensure that they are turned to silent.

I have the pleasure this morning of welcoming two new committee members. I welcome Mike Rumbles and Donald Cameron to the committee—we are glad to have you. I also put on record my thanks to Jamie Greene for his valuable contribution to the committee.

Before we move to our first agenda item, I ask Mike Rumbles and Donald Cameron whether they have any relevant interests to declare.

I have nothing to declare.

I have nothing to declare.

Thank you very much.

The first item on our agenda is an evidence session with officials from National Records of Scotland on preparations for the draft Census (Scotland) Order 2020. From NRS, we are joined by Amy Wilson, director of statistical services; Scott McEwen, head of policy, legislation and engagement; and Jill Morton, senior business lead, questions and collection instruments. They are accompanied by Scott Matheson, senior principal legal officer for the Scottish Government.

I invite Amy Wilson to make a short opening statement.

Thank you, convener. We are very happy to be here to discuss Scotland’s 2021 census to support the committee’s consideration of the draft census order and the accompanying documents.

National Records of Scotland has already been working over several years to develop a high-quality census, the next of which is planned for Sunday 21 March 2021. It will be the first one to be predominantly online, and it will meet the aspirations of society in 21st century Scotland. Hopefully, that will make the census more user friendly and provide choice.

The census is the only official count of every person and household in the country at the same time. It tells us who we are, how we live and how we work in Scotland. Essentially, it reflects the society in which we live. Scotland has relied on the information in the census for more than 200 years, and it remains the best way to gather much of the information that is required by Government, councils, the national health service and other users.

National Records of Scotland’s core purpose is to collect, preserve and produce information about Scotland’s people and history. We are very proud of our achievements through the census—the first of which was in 1801—and through all the other statistics that we produce, and we wish that to continue for 2021 and beyond. That includes ensuring that privacy is protected and that census records are held securely and confidentially.

Census outputs are essential to support decision making from national to local level, including allocating funding for schools, education, hospitals and infrastructure. Having accurate and reliable data is at the heart of the census. Billions of pounds of public funds are allocated in some way through that data, so it must be credible, and people must have confidence in it to take decisions.

As the committee is aware, the approach that we are taking with the census order follows on from the parliamentary committee recommendations from the 2011 census, where the committee asked the Scottish Government to simplify the procedure for future censuses.

That is why, for this census, we are starting the engagement with the committee early, with the aim of dealing with all the matters before the formal process begins. While the current process might thus be considered as the informal stage, please be assured that it is very much official engagement with the committee.

My letter of 5 September provided the committee with the draft census order and accompanying documents for discussion today, and for the committee’s consideration over the coming weeks. The draft order reflects the approach that we are proposing for the census in 2021, and the accompanying documents provide more information to support that, including the proposed guidance for respondents when completing the questionnaire.

Planning for the census is progressing well and our rehearsal is only one month away. Some matters—which will be highlighted through our discussion today—are still being finalised, and the rehearsal will provide excellent feedback on our readiness and on the approach that is being taken. Testing is on-going on many fronts, including on the sex question guidance, on which I provided an update with my letter. That is all being done to ensure that we deliver the best possible census to Scotland’s households in 2021, which will, in turn, provide the best possible data for our country. By asking the questions that reflect Scotland as it is today, we will ensure that the census continues to be a vital source of information for decades to come.

I end by reassuring the committee that National Records of Scotland will work closely with you to deliver the legislation that will allow for the census that we all wish to see for Scotland in 2021. The recent Census (Amendment) (Scotland) Bill demonstrated the interest that there is in census matters, and I thank you for your support through that process. I am in no doubt that such interest will continue through the census order process, and we look forward to working constructively with you to deliver Scotland’s 22nd national census.

Thank you very much for that comprehensive introduction.

The committee’s engagement on the census has obviously been shaped by our consideration of the bill. Over the past few days, we have had a lot of correspondence, principally around similar issues with regard to the sex question, which is what my initial questions will be about.

I want to ask about a few things. You can give yes or no answers if you want. The committee and the Parliament unanimously agreed to include a voluntary transgender status and history question in the census for the first time; I see that it appears in our draft questions paper. I should put on record the fact that I am very happy with that, too.

When you originally considered the census topic consultation back in 2016, you did not propose a transgender question—it came to you via stakeholders. Now that there is such a question in the census for the first time, would you say that that is a substantial win for the organisations that campaigned for a transgender question? A yes or no answer is fine.

It reflects the fact that there is a need for data, so it is a substantial win for data users, who will get the data that they need.

In its 2021 census topic research update in December 2018, the Office for National Statistics said of the gender identity topic:

“This is a developing area of research in many countries. Currently, no European country collects gender identity data in their census.”

That suggests that we are ahead of the curve in having a transgender identity question.

Yes, we are.

Do you think that any suggestion that the census is rolling back lesbian, gay, bisexual and transgender rights is a bit unfair, given that we have a transgender question in the census for the first time and we are ahead of the curve in Europe in doing so?

The transgender question is certainly a positive step forward, as is the proposed question on sexual orientation. There are obviously concerns on both sides about the sex question and how it is asked.

As you say in your sex and gender identity topic paper,

“Sex is a key demographic variable”,

and the sex question has been asked since 1801. You also say:

“There is a well-established user need for sex data ... It is a vital input to population, household and other demographic statistics which are used by central and local government to inform resource allocation ... and carry out service planning and delivery.”

However, you propose to keep the online guidance for the sex question that advises people to answer it according to how they feel, rather than with the sex that is recorded on their birth certificate. You specifically say that they

“don’t need a Gender Recognition Certificate”

to answer the question differently from their birth sex. We discussed that during the passage of the bill. Of course, that guidance, which you introduced in 2011, was only available online and people were not directed to it—it was quite difficult to find. Now there is a transgender question, so there is a question mark over whether the guidance is needed at all.

You will be aware that a number of leading statisticians and data users say that a self-identified sex question will damage data. We took evidence from Professor Susan McVie from the University of Edinburgh, who believed that it was a mistake. Yesterday, we received a letter from Professor Nick Bailey from the University of Glasgow, which was signed by seven other social science researchers—Dr Jo Ferrie, Professor Suzanne Fitzpatrick, Professor Christina Iannelli, Professor Sarah Johnsen, Professor Susan McVie, Professor Morag Treanor and Dr Beth Watts—from Edinburgh, Glasgow and Heriot-Watt universities. Their letter makes it clear that they support the transgender question and believe that it will be useful, but they believe that a self-identified sex question will inhibit their

“ability to monitor sex-based discrimination and disadvantage”.

How do you respond to the concerns of those eminent academics? Do you intend to engage with them?

Thank you for bringing that to our attention. That is the first time that we have heard it, as we have not been in touch with them and they have not been in touch with us. We will engage with them—absolutely.

There have been suggestions that what was done in 2011 damaged the quality of the data. We went through an extensive quality assurance process involving quite a lot of academics, people working in local areas and expert statisticians. At that point, there was no evidence that we could see anything in the data that suggested that we had introduced something different from what had happened in previous censuses.

However, it is obviously a concern, so we will continue to engage with people. This time round, we have recognised that we do not understand enough about the effect of providing the guidance and people using it. That is why we are doing testing to understand how much either not looking at the guidance or having different versions of the guidance affects how people respond to the question. It will be only when we have the information from that testing that we will fully understand whether different versions of the guidance lead to different responses.

You are going into an area that we could have a whole separate committee meeting on. Many people would argue that women are discriminated against because of their biology, but that is a whole separate area.

The 2011 guidance was online guidance and my understanding is that only about 20 per cent of people answered the 2011 census online. Is that correct?

Yes.

Is it also correct that the online guidance was not necessarily flagged up—it was not beside the question and people had to seek it out?

That is correct.

It is therefore fair to say that very few people would have been aware of that online guidance; in fact, it has been drawn to our attention that Stonewall’s submission to your census topic consultation in 2016 suggested that it was unaware of the online guidance because, according to it, sex and gender are different things and people would not know how to answer the sex question. The main LGBT charity did not seem to be aware of the guidance, and when you engaged with stakeholders for the 2021 census, the feedback from trans people indicated that many of them did not seem to be aware that they could answer the sex question in the way that I have described. It seems to me that even your own material shows that not many people were aware of the guidance, so it would not have affected the data in 2011.

That is a fair point. I do not think that we know how it affected the data in 2011. From looking at the data and the quality assurance that we have done, there is no evidence to suggest that we started to see trends that were different from anything that had happened in the past. However, you are right—we do not know how the guidance affected people and we do not know how many people actually looked at it in 2011.

It was different then; the issue is now much more high profile. Society has changed quite considerably, and I believe that that is why the social scientists I mentioned are so concerned about changes to the guidance.

I know that other members are going to ask about the on-going testing of the guidance. Correct me if I am wrong, but I understand that although you plan to recruit 5,000 people randomly, you will find members of the trans community through trans organisations to test the guidance.

Yes, that is correct, but I think that we will be recruiting more than 5,000 people. I will ask my colleague Jill Morton to answer that, because she knows the details.

With the methodology agreed, you are right that, when we had the stakeholder events around the testing, our figure was around 5,000. We are now targeting 6,500 households for the general population. We are advertising through a broad variety of organisations that might touch on the trans community, so it is not necessarily just—

You will recruit the trans respondents through trans organisations. The committee is very aware that those organisations have strong views about the subject. You will recruit people for a testing exercise through organisations that already have very strong views about the questions. Is that not liable to damage how you collect the data and the outcomes of the testing?

We will recruit the trans respondents partly through those organisations, but we are being very open in how we recruit. For example, we have been in touch with the NHS, and we have looked at whether gender clinics can help. We have been open in saying to all stakeholders that we are keen to hear from anybody who wishes to be involved in the testing. You are absolutely right to say that we want to have a broad range of people from the trans community, so that we can hear as broad a range of voices as possible. Although working with the stakeholder groups has been the biggest way in which we have recruited, we are working on other ways, too. As I said, we wish to ensure that we hear a broad range of voices.

I have some questions on the two versions of the proposed guidance that you are consulting on. The key issue for me is the interaction between the sex question and the voluntary trans questions. I am not clear how that works. For example, the guidance that seems to be closest to the 2011 guidance, under the heading “How do I answer this question?”, says:

“If you are transgender the answer you give can be different from what is on your birth certificate.”

It says “can be”, so the person has a choice. If someone decides to answer the voluntary trans question, I cannot see how that will give enough evidence on the answer to the sex question. How will the users of the data work out how the questions interact, if they want to work that out? I am not sure how there will be consistency in the way in which the questions interact. It seems that a lot of choice and flexibility is involved.

The interaction in that case is specifically around how respondents approach the questions and understand what they are asking. We have done quite a lot of testing on that. Across all the testing that we have done, we found that different people approach the sex question with different understandings. The majority of the population do not distinguish between any of the definitions—for want of a better word—for the sex question that we have looked at during census development. Having a trans status or history question directly after the sex question changes how some members of the community approach the sex question. Similarly, being able to see the sexual orientation question changes the way in which people approach answering those three questions. It is about how people answer the questions, rather than about the interaction for data users.

It is not for data users. You are not directing people by saying what they are meant to do. You say that their answer “can be different”; it is up to them. I do not see how that will help to collect accurate data, or what information data users will be able to take from that, apart from, “We don’t really know how this person is answering.” I think that, next week, you will provide the opportunity for us to see how the form operates, so that might help my understanding.

There are also two choices for the advice that is given to people who identify as non-binary. The proposed guidance says:

“If you are non-binary or you are not sure how to answer, you could use the sex registered on your official documents, such as your passport or driving licence.”

It then says that there is also a question about trans. The other advice does not give as much detail. I do not know why you have chosen two different options for the guidance for people who are non-binary. Why is there different advice?

Is your question about why we have not made comments about non-binary people in the guidance that suggests that people should respond with their sex as registered on their birth certificate?

Yes. You give two options. One option says:

“If you are non-binary or you are not sure how to answer”

the sex question,

“you could use the sex registered on your official documents, such as your passport or driving licence”.

It then says that the next question is about trans status and history. The alternative proposed guidance that has been tested says:

“The next question is a question about trans status and history. You can respond as non-binary in this question.”

It look as though that guidance does not give any advice to non-binary people on how they should answer the sex question. Why is that?

The guidance in that second version is that they should answer with the birth as recorded on their birth certificate. That would mean that they could not be non-binary so they would have to respond as either “Male” or “Female” as on their birth certificate or, if they have a gender recognition certificate, as what is on that.

We worked on the guidance with stakeholders from women’s groups and equality groups to make sure that what we were going to test would make sense to people, and what came back from the stakeholders was that there was no need to provide any information for non-binary people in that guidance because it would not be applicable.

So the first guidance suggests to non-binary people that they can answer it as they have identified themselves on a passport or driving licence, and, although it does not say it, the second one expects people to register as what is recorded on their birth certificate. The second suggested guidance does not tell a non-binary person what to do. Rather than trying to explain that to me today, you say that you are testing the guidance. When will we get further information on what works?

Also, you said to the convener that you are doing a test of the guidance. What weight is put on the testing that you do? Is it just that the proposal that gets the most responses will be the one that you use, or will other factors influence the decision?

It is a complex issue. It is certainly not going to be straightforward. We need to understand whether people answer differently.

We are testing several things. First, we will ask people to answer the question without having any guidance. It is a split sample, so half of the population will get one set of guidance and half will get the other set. We also want to understand how people answer when there is no guidance, and how they answer when they get whichever set of guidance they are given. We are also asking questions about how acceptable it is and whether, having read the guidance, it would change their response. We did not understand in 2011 how the guidance affected how people responded.

All those things will need to be looked at if we are to understand whether the question is acceptable to people, whether the guidance changes the response and, depending on which version of the guidance we use, what we are actually measuring. I will let Jill Morton answer the question about timescales, but we will have to look at all those issues and report back to the committee so that members can understand what we found.

We are looking towards mid-December to the end of December as a date for the completion of testing. These things do not necessarily go according to plan, which is why there is a little bit of flexibility. It will be before Christmas.

Convener, I have some other questions. Shall I leave those to let other people in?

We will do them once we have got most of this topic out of the way.

I thank the witnesses for coming in to continue our discussions on this important subject.

I am a lawyer by trade so I cannot help looking at this from an a priori perspective. In the latest version of the proposed questions, question 4 is:

“Do you consider yourself to be trans, or have a trans history?”

Under question 4, it says, explicitly:

“Trans is a term used to describe people whose gender is not the same as the sex they were registered at birth”.

Preceding question 4 is question 3, which is “What is your sex?” and the choices are “Female” and “Male”. In the interest of consistency, how could you possibly have a definition of sex that is other than sex registered at birth? At question 4, your key point is about the sex as registered at the person’s birth. How could you issue guidance for question 3 that is not in accordance with the position as stated in question 4? From a legal perspective, I just do not get that; I simply do not understand it.

This is not entirely a legal point, and I am not entirely sure that I understand the premise of the member’s question, convener. Is the member talking about the second option of the guidance, which talks about the answer having to be the same as what is on the person’s birth certificate and then goes on to talk about a gender recognition certificate? A gender recognition certificate will have the effect of changing what is on one’s birth certificate, so I am not sure whether the member is suggesting that the legal effect of the Gender Recognition Act 2004 should be disregarded entirely.

No. With respect, I am not suggesting that at all. As you say, the legal position is that a gender recognition certificate would supplant any prior document. That is well understood; that is the axiomatic legal position. Rather, my question is this: how could you possibly issue guidance in accordance with the current language of question 4 that suggests that sex could be defined as anything other than sex registered at birth? I just do not understand how, from a legal perspective, you can have such an apparent lack of consistency between questions 3 and 4.

With respect, I think that the member has just said that sex registered at birth must be the only definition of sex that can be relevant, but—

That is the language that the NRS uses at the moment. It is understood that the gender recognition certificate would, as a matter of law, supplant a previous document, but you are talking about a person’s sex registered at birth being the key element in question 4 in terms of how you define trans people. My question is very simple: how could question 3—that is, “What is your sex?”—have guidance that suggests anything other than that that is a person’s sex registered at birth? Clearly, the gender recognition certificate is a bit of a red herring, because the legal position of that is well understood.

Maybe the NRS wants to reflect on that point further. As a matter of law, I think that the draft is fundamentally inconsistent. The census is to be a gold standard of statistical data collection. I do not see how it will obtain that standard as it is drafted. As a matter of law, it does not really work for me.

Perhaps I can help here. Obviously, the effect of a gender recognition certificate is to change a birth certificate. Therefore, a person with a gender recognition certificate can change their birth certificate. That is completely different from what you are proposing in the sex question, which is that people can answer that question even if they do not have a gender recognition certificate that changes their birth certificate.

You could say that there are two classes of people. There are people with gender recognition certificates who have changed their birth certificate and there are people who consider themselves to be transgender who do not have any legal documentation. You are saying that the latter should answer the question how they like. That is the critical issue.

Surely it would be easier and much simpler if question 3 asked, “What was your sex registered at birth?” That would lead on to the wording in question 4.

We could ask that question, but it would need to be understood that that would be likely to lead to quite different data from what we have asked for in the past. Again, that would be asking a more specific question than the one that we are currently proposing—that is, “What is your sex?”

If we were to ask Kenny Gibson’s question, we would need to understand exactly what the effect of that would be. I think that we are reflecting data users’ needs. Yes, some data users would probably say that what you propose is what they need, but other data users would not say that and it would not meet their needs. We are trying to meet a broad range of needs through the sex question.

I will make a point about the underlying objective, because it is always important to go back to first principles. The objective is to ensure that how we collect our data, the methodology that we use and so on are to a gold standard. Consistency is important, too.

In that regard, we can go back to questions that we asked when considering the Census (Amendment) (Scotland) Bill. This is not about how a person feels about a particular question—there may be lots of other questions in the census that lots of individuals find slightly intrusive for whatever reason. This is a question of collecting data for the benefit of the state on all aspects of public life, public services provision and so on. I would have thought that the key objective is to ensure that, for data users, this is the best that we can do. Other matters can be considered, but that should be the key consideration. That does not seem to be what you said in your previous statement.

I am sorry if I was not clear. I was trying to say that the data does not have one set of uses. For some people, the data use may be biological—about sex in a legal sense—whereas for others, it may be more broadly about other people’s perceptions and discrimination. In all our work, there was not a single use case in terms of what the data is needed for. A question about sex registered at birth may meet the needs of some data users but not those of others—that is part of the issue.

That issue was raised by Mr Gibson. My key point remains that there is a legal inconsistency between questions 3 and 4.

My last question is about all our discourse about the Census (Amendment) (Scotland) Bill and where we are with the pre-draft of the census order. A lot of this discussion could have been streamlined if the NRS had had early engagement with the statisticians. I do not understand what the process has been to have a plea from statisticians in an email on 11 September at 10 past 2 in the afternoon saying that they have serious concerns. How have we got to this stage? Could you enlighten the committee?

We have spent a lot of time consulting people, particularly over the past nine months. We have sent out extremely clear updates to discuss what we were doing with thousands of people who subscribed to our newsletters. Many of the statisticians will be linked to groups to whom we have given presentations and talks. We will follow up on the email, but its senders have not come to us or engaged in the past, nor were we aware that there were any concerns, despite being as open as possible about what we were doing over the past nine months and before.

They will speak for themselves as to their perception of the engagement. Thank you, convener.

What engagement have you had with Professor McVie?

Professor McVie is a member of the board for official statistics, as you know. We have had discussions, and we presented at the most recent board meeting. She and I have had several conversations about her views on this matter.

Thank you.

I will pick up on a suggestion that was made a moment ago. If question 3 were to ask about a person’s sex registered at birth, how would that interact with protection for individuals with a gender recognition certificate? Can a person with the certificate be asked that question?

I will defer to my legal colleague.

Ross Greer will no doubt be aware of the legal protections of confidentiality around gender recognition certificates—I take it that those issues are what he is getting at. The legislation that we are dealing with is the Census Act 1920; the order and regulations that are made under it would be enactments that the Gender Recognition Act 2004 acknowledges as exceptions to the confidentiality provisions. The disclosure on the census form would go through the process in accordance with the enactment. I think that the exception is in section 22 of the 2004 act.

That is the legal position, and more generally there is a policy position about the reason for those confidentiality provisions in the first place. There is an overall balancing exercise; private rights of privacy in particular circumstances are balanced against the public need for the outputs that are generated by the census.

Thank you. In response to the convener’s original questions about the 2011 census and the guidance around it, could you clarify the situation with regard to the sex question before the 2011 census?

In previous censuses before 2011, we never issued any guidance. We have always asked a question that has had the response options “Male” or “Female”. The question has been asked in a slightly different way across the years, but it has been in essence the same question, with the options “Male” or “Female” and with no guidance.

What is your best understanding of how people were answering the sex question before 2011?

We do not have an understanding of that, because we never tested that, and we never had any guidance. In the past, there has almost de facto been self-identification. I do not necessarily mean that it was how people would interpret the question today. However, it is a self-completion exercise, so we get back what people have said, based on how they interpret the question. We have looked back through the records since 1801 and found no record of testing being done around that question to understand how people interpreted it.

All of this—by which I mean the discussion and the huge amount of correspondence that we have all received over the past couple of months—leads me to ask the question: is guidance on the sex question required? What would be the effect of going back to the pre-2011 position of not providing guidance? I say that from the understanding that it will be a challenge for the Parliament to come to a unanimous view on a definition for guidance. I believe that the question of whether guidance is required has been raised with you by those on both sides of the debate—to reduce this to a binary, ironically enough.

From the testing, we will get an understanding of how people respond without guidance, whether they can respond and how the response changes with guidance. If I recall correctly, during the committee’s evidence sessions on the Census (Amendment) (Scotland) Bill, Professor McVie raised the point that, if we do not have guidance, we do not understand accurately what we are measuring. Given everything that has been said, the concern is that, by not having guidance, we might not have any understanding of what we are measuring. However, the intention of the testing is to understand whether having guidance actually changes people’s responses to the question and how those responses change depending on which version of the guidance we use.

What is your understanding of how other census models in the English-speaking world approach the question?

There are various approaches. Frankly, countries are all doing different things. Some census offices are looking at non-binary responses and some are considering asking a question around sex at birth and how people currently identify. In England and Wales, the ONS published guidance yesterday that says that, for 2021, it will continue with what it has done in the past and will have a binary question. It has said that people who are trans or non-binary do not need to answer with the sex that is on their birth certificate.

I presume that the ONS has tested and come to that conclusion on that basis. Do you take the work that the ONS does into account, given that Scotland and England have the most broadly comparable systems?

Absolutely. We work closely with our colleagues at the ONS. We have drawn strength from some of its testing, and vice versa. Many of our users, as well as the office for statistics regulation, have pointed out that harmonisation across the UK is vital on such a key demographic variable.

I have a supplementary question, just for clarification. It may have been answered already, so I apologise if I am asking a question that has already been answered. Does the NRS testing also involve a no-guidance scenario?

It is not specifically for a no-guidance scenario, but the first part of the testing asks people to answer the question without any guidance, so we will see how they answer in that situation. Some of the follow-up questions ask whether the guidance changed the answer to the question and ask about acceptability.

Regarding the other countries that you talk to and look at, have you been in dialogue with the likes of Australia, New Zealand and Canada on the questions that they have posed?

We absolutely are in dialogue with those countries. In fact, colleagues of mine were at the international census forum last week, which was held in Ireland and was attended by colleagues from those countries. The issue was one of the big points that was discussed. If you like, we could write to the committee to provide our best understanding of what those countries are going to do. Jill Morton was there, so she might want to say more about what was discussed last week.

There was a lot of discussion about that because, internationally, there is no clear single question that works for every country. We are all working within our own political and social contexts, which help to form the questions that can and cannot be included. When the testing is complete, we will share our results with the international census forum, because the area has been picked up internationally as one in which there is a bit of an evidence gap, and we need to understand whether, if the guidance is changed, that changes how people answer the question.

There has been a suggestion from other countries that the introduction of a legal or biological sex question in Scotland would go against international practice and, if Scotland were to do that, the approach would be somewhat different from what it is elsewhere.

As I mentioned in my previous response, in the international census community, there is no single question that suits all countries, as we all have slightly different needs. In Scotland, we work to our data users’ needs with the clear awareness that we have a UK responsibility and we must be aligned. Other countries also have to meet their data users’ needs.

We and the ONS have spent a lot of time talking to data users about the issue. The trans status or history question, for example, has been developed for the Scottish context, but the same primary purpose applies—we must meet our data users’ needs. Other countries need to do the same in their own contexts, so, although we work closely together and learn from one another, we have not got to a position where we have one question that everybody should ask.

I am struggling to understand why guidance is needed for the question “What was your sex as registered at birth?” because the answer is either “Male” or “Female”. Why would guidance be needed? A person would tick one of those two boxes, because they will have been born either male or female.

I would agree if we were to ask that question—and it probably would be easier if we did. However, that is what we are testing at the moment, because that is not the question that we ask now. We ask, “What is your sex?” and we do not define in the question what we mean by “sex”. That is why guidance was introduced for the previous census and why, this time, we are testing guidance on the interpretation of the question “What is your sex?”

Would it not be clearer and more helpful to make the question “What was your sex as registered at birth?” rather than go through elaborate guidance to see how many camels can fit through the eye of a needle or dance on the head of a pin, or whatever the saying is? It seems that we are making things incredibly complex when they should be relatively straightforward.

As I said in my previous answer, if we were to ask that question, it would meet some data users’ needs, but some data users would feel that it definitely did not meet their needs. It can be considered, but it is different from the one that we have asked over the past 220 years. Therefore, I suspect that we would need to understand the effect that changing it would have on long-term time series and comparability around the UK.

From what you have said, it sounds as though no question would meet everybody’s data needs. You said that the question has not been in the census in the past 200 years, but people did not really think much about things such as transgenderism in 1801, 1901 or even 1951. It was never something that I came across at all when I was growing up, and it has only been fairly recently—in the past few years—that I have. It is quite difficult to suggest that, because it was not in the census at the time of the Boer war, we should not ask it now. It is surely about trying to get the data that we need in the clearest and most straightforward way possible. Asking someone the question “What was your sex as registered at birth?” could not be simpler or more straightforward.

I am not suggesting that we should not ask the question because it has not been asked in the past; what I am saying is that, if we were to ask that question, it is a different question, and we would need to understand what the effect of asking it is, because it is not necessarily the case that asking that question would give comparable data to, for example, the data that we got in 2011 or the data that you would get from asking “What is your sex?” That is all that I am suggesting.

Are there any other questions on that area of the census?

Yes. On the issue of guidance, how many other questions have got guidance attached to them?

All the questions have guidance attached to them. We have provided the committee with the draft guidance for all the questions.

And that guidance will be published online along with the census.

Yes, it will be.

Is that a new thing? My understanding is that the guidance on the sex question is a new thing. It was introduced in 2011 for the first time and, if you look at the period from 1801 to today, you will see that that is a statistical blip—it is the oddity, not the norm. Was all the other guidance published online in 2011, even if, at that time, it was not an online census?

Yes, it was. Again, we are trying to provide people with as much clarity as possible. Obviously, there are different levels of guidance. I hope that, when we can show you the online system, you will see that, in some cases, there is guidance in the question, in a way. That is the case with the trans question, which you talked about earlier.

People can get different levels of guidance. They can get quick top-line guidance that tells them what the question is asking about but, if they want to delve down further, much more information will be available.

A moment ago, you talked about the need for harmonisation across the UK. Are you suggesting that every question is identical to a question south of the border?

No—absolutely not. However, equally, there is an agreement across the UK countries on how we conduct our censuses. We want to strive for as much harmonisation as possible, but we recognise that, ultimately, censuses need to deliver what the data users in the country need. Harmonisation is relevant in relation to issues such as sex, which feeds into many decisions, including funding decisions. Those things are important, so that is one element that is considered.

I have some questions in that area before we move on to questions from other members. It is best to deal with these issues now rather than coming back to them at the end.

As Annabelle Ewing and others have suggested, the 2011 guidance was something of a blip, and not many people knew about it, even within the trans community. One of the concerns that people have about going back to just asking “What is your sex?” without any guidance, as has been done since 1801, relates to the issue of confidentiality. I do not know how people are going to answer the question. Can you offer reassurance on the confidentiality of the census? My understanding is that it is a criminal offence to unlawfully disclose census data and that a person who does so may be fined up to £10,000 or sent to prison. It is incredibly secure, is it not?

It is incredibly secure. We work hard to ensure that that is the case.

That should reassure people that, really, no one is going to challenge how they answer the sex question. No one is going to come back and say that they have answered wrongly, because it is completely secure.

That is absolutely the case. Further, we do not check its veracity against any other information.

That might reassure people on that point. No matter what conclusions we come to here, the answers are secure.

I said that we were not going to discuss this in the committee, but I would like to come back to the point that you raised initially about the issue of sex discrimination and biological sex versus perceived sex. The Equality Act 2010 has nine protected characteristics. One of them is sex; another is gender reassignment. That legislation therefore contains an understanding that those two things are different, and the guidance notes for that act make it clear that gender reassignment is a different thing from sex. I take it that you are conversant with the 2010 act.

Yes—absolutely.

It seems to me that you are slightly jumping ahead of the law. We are having a debate across the UK about changes to the Gender Recognition Act 2004 and whether we should move to self-ID. By making the sex question in the census one of self-ID, you seem to be jumping ahead of that debate. We have not passed legislation on the matter in either Scotland or the UK; indeed, the Scottish Government is reconsulting on it because it is the subject of a debate. Why are you jumping ahead of that in the census and suggesting that the sex question will be a matter of self-ID and not one of legal sex? I think that most of the people who object to what you are doing are quite happy to compromise and say that they are happy for people to answer on the basis of their legal sex if they have a gender recognition certificate, because the number of those people is so small that it will not make any statistical difference, but you are jumping ahead of that and saying that people can self-ID their sex. That seems to me to be problematic.

I will ask Scott Matheson to speak to that, but I think that that comes to the issue of what sex is as defined in the Census Act 1920. Scott, do you have anything to say on the Equality Act 2010 and what we are actually asking?

The starting point would be that the Equality Act 2010 does its job and the Census Act 1920 does its job. There is a connection in that the data that is gathered in the census will feed in and be used by a number of different data users—Amy Wilson and my other colleagues will be able to say a bit more about that, if necessary. Therefore, there is a link, but legislation has to be seen in the context of the job that it does. The Equality Act 2010 gives rights and duties and governs what behaviour is acceptable and how people should be treated. A fundamental principle that runs through it is the dignity of the individual. It will therefore be applied in the circumstances of particular cases.

Individuals will be legally of one sex either because they were registered as that sex and they have not gone through a gender recognition certificate application process and got a gender recognition certificate, or because they have such a certificate and therefore have the sex of their “acquired gender”, to use the language of the Gender Recognition Act 2004. The interplay of the two protected characteristics that were mentioned—those of sex, taking the definition that it is given in the Equality Act 2010, and gender reassignment—is complicated, and that has to be looked at in the context of the particular case. The census does not go down and deal with things at that micro level; it does not arbitrate between the rights and obligations of particular parties in particular circumstances. It generates data for a wide range of needs.

Arguably, it would be wrong to completely hitch the census to the wagon of the Equality Act 2010, because the definition in that act does its job with all the things that are round about it. It deals with direct and indirect discrimination and with discrimination based on whether somebody has a protected characteristic or is perceived as having that protected characteristic. All those things are wrapped up there. The census has to do something entirely different: it has to collect data and address a wide range of user needs.

At one level, it might seem superficially attractive to copy and paste the language of the Equality Act 2010 into the census but, if we did that, the census would not do the job that we ask it to do.

I am not suggesting for a minute that you should do that. What I am suggesting is that Amy Wilson’s comments at the beginning attempted to define sex as something that is not biological, but a matter of self-identification. You say that you are not going to cut and paste the Equality Act 2010 into the census but, on the other hand, you are imposing a particular view on the census, which is that sex is something that people can self-identify. That does not exist anywhere in law.

The Equality Act 2010 does not define sex in strictly biological terms. It must admit of the possibility of somebody possessing a gender recognition certificate and therefore, at a minimum, it deals with legal sex rather than strictly biological sex. So—

The point has already been made to you that people are willing to compromise on that and to accept that people can have a legal sex, but you go beyond the Equality Act 2010 and the GRA and suggest that sex is a matter of self-identification.

That takes us back to what we are trying to get out of the census and the user data that we are being asked to generate.

Users who have problems with the issue have pointed out that, with subsections of the population—not the general population but subsections of it as society changes—if we change the way that we define sex, that will affect the data. That is why we have letters from data users and social scientists who say that there is an issue, particularly in the area of defining sex-based discrimination. The reason why sex is a protected characteristic in the Equality Act 2010 is that there is a recognition of such discrimination.

Yes, but I have tried to explain that the Equality Act 2010 definition sits in its own context and within the framework of the other things that are around it. That does not give a perfect answer. If we had something that absolutely guaranteed that people would answer the sex question according to whether they were actually a man or a woman in the sense that the Equality Act 2010 would have it, that would not necessarily tell us the most useful thing that we need to know about those individuals in generating the data for a range of user needs. The data that are gathered are not being used to determine whether an individual is being treated less favourably because of the box that they have ticked.

We have a letter from nine professors and doctors, led by Professor Bailey, who would disagree with you on that point.

Donald Cameron has a supplementary, so I will bring him in.

I want to pursue the convener’s point on the Equality Act 2010. I accept Mr Matheson’s point that the Census Act 1920 and the Equality Act 2010 have different purposes, but the Equality Act 2010 is the seminal and definitive piece of UK-wide equalities legislation, and it combines everything together in that regard for the first time. The definitions in the 2010 act may require to be revisited, but surely, as the convener suggests, there is a strong argument for, at the very least, consistency between that legislation and the legislation that we are considering.

There may be, but that is a much more wide-ranging policy question than I can speak to. The member might be suggesting that, when the Equality Act 2010 was passed, consequential amendment should have been made to the Census Act 1920. I am not sure that that was in Parliament’s mind when it passed the Equality Act 2010.

In all our endeavours, whether in relation to the census or anything else, should we not be required to operate in the context of the Equality Act 2010? The argument about amending the 1920 act is therefore a bit of a red herring. As legislators, we have to operate within the strictures of the 2010 act in all our endeavours, irrespective of the area of activity.

There is a question in my mind: in what way does what is being proposed in the census order not fit with the Equality Act 2010? The act is not being modified and nobody’s rights or duties will be changed as a result of what is being done in the census. There is no suggestion that what is being done relates to reserved matters in the act, or that it would be outwith the powers of this Parliament to make the provision that is proposed in the draft order.

As Amy Wilson has indicated, a wide range of data users look to the census to get that data. It does not necessarily follow that what they need and therefore what should be generated to support that wide spectrum of needs has to follow slavishly the terminology or the meaning of the Equality Act 2010.

That is quite a bold statement.

It is quite bold. The Equality Act 2010 runs right through the documentation—when you define sex, you mention that sex is a protected characteristic in the act, and you conduct an equality impact assessment because the act demands that of you. The act is at the core of what you do.

I am sorry if we have perhaps been unclear if we have suggested that the 2010 act is not important, or, indeed, central to aspects of why the sex data is needed. However, that is different from saying that there can be only one right answer to what must be asked in the sex question and that the answer must be what someone’s sex is as understood in the 2010 act.

You have said that you will engage with the people who have written to you. The statisticians’ point is that the census affects lots of other statistical exercises. In their email to the committee, they say:

“It is the only source providing full coverage of the population on such a wide range of aspects of social life and therefore it is uniquely well-placed to provide information on smaller population groups.”

Society is changing. In 10 years’ time, people could be defining themselves in all sorts of ways, and the statisticians are really concerned that the proposed approach will damage data.

I understand that you are under an obligation to the UK Statistics Authority to engage with statisticians, and that it could intervene if you do not do that. I take it that you will be engaging with them.

My final point is on stakeholder engagement. You are engaging stakeholders on the equality impact assessment. Will that take place next week?

That is correct.

I have been approached by members of stakeholder groups—not the professional, publicly funded stakeholder groups, but the independent ones that you have been engaging with, which were given one week’s notice of the date. They were originally told that they could meet in Glasgow or Edinburgh, but they have now been told that they can meet only in Edinburgh. Many of them will not be able to make the event because they have had such short notice. Will you look at that, given that some people are unable to attend?

Absolutely. We would be very keen to look at that because we are keen to get feedback on the equality impact assessment and the other assessments. We want to ensure that they represent the broad range of evidence, that we are not missing things and that they do not contain things that do not represent people’s experience.

Thank you very much. We now move on to other aspects of the census.

You have talked about the census primarily being online. You have also talked about your planned rehearsal—I think that you said that it will take place in a month’s time. How will you progress the rehearsal? What safeguards are being put in place? I understand that you will be using different technology for the census. Will you be using it all during the rehearsal? If there are risks, how will you mitigate any difficulties that arise from the process?

The rehearsal is due to take place on 13 October in three local authorities and will involve 72,000 households. We have been going through a very long testing process to make sure that we can test for security issues, such as denial-of-service attacks, and what to do if systems go down.

Security is very important to us. People are trusting us with their data and we have to keep it safe.

The results of an independent assessment of all three census offices have been published, highlighting areas where we are good and those where we need to do more. We are going through a lot of security tests to make sure that the information that we collect from people will be safe. We have published a lot of information, including the data protection impact assessment, in advance of the rehearsal to reassure people about what we have done.

Do the guidelines that you have set out for the rehearsal across the authorities that you have chosen say that they will feed back to you regularly? When you have completed the rehearsal, how will the end of the process be managed so that the information comes back and everyone feels that it has gone securely? The biggest concern is that there could be issues with the majority of the census being online. It is difficult to make sure that every aspect is covered. I appreciate that you have done some work on that, but it comes down to the guidelines that you are putting out and how you will manage that process.

A crucial reason for the rehearsal is to test on a smaller scale so that we that we have all the processes in place, including those to deal with incidents, should there be any. We will do a comprehensive report after the rehearsal, which will be shared publicly because we want to be clear about what we have learnt from the rehearsal—what has gone well and what has not gone so well. We have policies that have to be followed; if there are any information security breaches, we will make sure that we follow all the guidelines.

If anything crashes or there is a major issue in the process, how will you revise the process?

If there were to be a major issue, we would want to make sure that it was rectified. We would identify the issue in our report and be very clear about the steps that we would put in place to make sure that it was tested to provide assurance before 2021.

I will move on to the questions about ethnicity, which I understand to be question 22, on national identity, and question 23, on ethnicity. I assume that you want to be as accurate as possible, to get all the information that you need, and accuracy needs consistency. After question 23, “What is your ethnic group?”, section A is based on a colour; sections C and D are based on geography; section E is based on geography and colour; and section F has religion in it. Is that consistent?

Our current ethnicity classification has had an issue for some time around the use of colour terminology and geographic information. I will come back to your point about religion.

A lot of work before 2011 took us to a certain point, and work has been done with stakeholder groups since then. Some stakeholders do not like the use of the term “Black”, and we moved on in 2011 by separating out the “African” category. We consulted on whether the term “Black” should be removed this time around. I will ask Jill Morton to respond to your question on that. We had a very low response the last time around, and we wanted to know how people would be able to respond if that category were to be removed. We also wanted consistency over time.

I accept that the question involves different concepts. For 2011, we tried an ethnicity framework that was entirely geography based, but that did not work very well either. There is an on-going process with stakeholders to get information that allows people to address the policies that they want to address and to respond to issues but that maintains consistency over time. One comment that we got back from a lot of stakeholders was that, given that this is such a key variable, if we completely changed the options, it would be difficult to see how the situation had changed between 2011 and 2021.

I understand that.

In the consultation on the use of the term “Black”, we talked to a small number of groups because, particularly in 2011, there was a small community that identified in that way. We had some evidence that we should keep the terminology consistent with that which was used in 2011, while other stakeholders suggested that the terminology needs to be changed. The stakeholders who are data users suggested that they need the terminology that was used in the question in 2011, for reasons of consistency.

The bigger point about the NRS consultation on the issue is that, from a data use point of view, we were not presented with a strong need to change the terminology; we were presented with a need to retain what was done in 2011.

You seem to be saying that you want to be consistently inconsistent, because the questions are based on geography or colour. I will give an example of what I mean. Question 22, which is about national identity, is an easy one. It says:

“What do you feel is your national identity? Tick all that apply”.

As a layperson—I am new to this topic—I would tick “Scottish” and “British”, because I feel both. The next question, which is 23, asks:

“What is your ethnic group? Choose ONE section from A to F, then tick ONE box which best describes your ethnic group or background”.

The first section is “White”, and it has the options “Scottish” or “Other British”. In my case, what is the difference ethnically between being Scottish and British, or English or Welsh? What is the difference ethnically?

The Scottish Government is one of our key stakeholders, as a key user of the data. In the format of the question, “Scottish” is separated out to meet certain needs. We have not been presented with a strong need to separate out, for example, English, Northern Irish or Welsh in the question. The ethnic group question is one of the most widely used questions in the census. It is another question that is derived in a way to meet a broad range of user needs. I accept your point entirely—

I just do not know how to answer the question—it is as simple as that. If you are after accuracy, I do not think that you are going to get it.

A lot of work has been done on the ethnic group question and the related concept of national identity, as well as religion, which comes into the area as well. They are all slightly different facets of how we identify as individuals. The question is complicated and the topic is a sensitive one for a lot of people. We have presented the question in that way to meet a number of data users’ needs, and the testing that we have done for those purposes shows that it meets those needs.

I will finish by making the point that I am confused. In my view, the question is too complicated. It asks a question about ethnicity that is not actually ethnic. The question on ethnicity confuses colour and geography and mixes everything in. From my perspective—this has also been put to me by others, which is why I am raising the issue—that is not consistent. If you want accurate information, you have to ask the right question.

I have a question about census questions 21 and 23. For question 21, which is the question on religion, the possible answers include “None”, “Church of Scotland”, “Roman Catholic”, “Other Christian, please write in below” and “Muslim, please write in below”. Should that not be clarified to say that people have to write in which denomination of Islam they are? Unless specifically asked for the denomination—whether that is Shia, Sunni or whatever—somebody might just write “Muslim” in the space, as well as ticking the box.

On the digital platform, the suggestion that you have made is how it is presented. The version of the questions that has been supplied to the committee is derived from the paper questionnaire, in which there are greater limitations of space and design. Next week, there will be a demonstration of the online version.

That is okay—it was just for clarification.

Under question 21, which is on religion, a possible answer is “Jewish”. However, under question 23, which is on ethnicity, Jewish people have also been classed as an ethnic group. If I was to convert to Judaism tomorrow, would I tick “Scottish” or “Jewish” under question 23? There is surely a wee contradiction there, because one best describes ethnic group or background but, if you convert, you might feel fervently about that. Why is “Jewish” a possible answer for ethnicity and religion?

We have worked with stakeholders for a long time on that question. The Jewish and Sikh communities—at least, certain parts of those communities—have strong views about the fact that Judaism and Sikhism are both religions and ethnic backgrounds.

There is a strong sense from the Jewish community and some evidence from 2001—when we asked a question about religion of upbringing as well as current religion—that asking a question such as question 21 undercounts the whole Jewish population, and people who are not practising Jews but are ethnically Jewish and might want Jewish end-of-life services or other such things do not get counted.

Sikh communities made similar points, so we worked with Jewish and Sikh communities to see whether we should include “Sikh” and “Jewish” in the ethnic group question as well as the religion question. There were split views on that.

We did some survey work with the Jewish population and found that, because of all sorts of connotations, some people were not at all keen on having “Jewish” listed as an ethnicity, but some people were keen. We looked at having a separate tick box for “Jewish” under the ethnicity question, but it did not test well and some people were very uncomfortable about that. However, working with Jewish groups, we found that having the tick box as a prompt was the best way to indicate to those people who deemed their ethnicity to be Jewish that, whether or not they ticked “Jewish” for question 21, it was acceptable to do so.

With the Sikh community, it was slightly different. When we looked at having a “Sikh” tick box or prompt, we found that it was not considered as acceptable, and we have worked with the community to agree that we will not have “Sikh” as a possible answer to the ethnic group question.

Barbra Streisand is Jewish, and so was the late Sammy Davis Jr, but nobody would class them as having the same ethnicity. Does this mean that, in the future, the census will list “Muslim” as an ethnic group? Will we end up merging questions 21 and 23? There is an inconsistency in some religions being classed as ethnic groups and others not. Many religions have a strong ethnic base.

The Jewish and Sikh communities are the two groups that have approached us about the issue. In addition, it is my understanding that there have been legal cases in which it has been determined that those religions are ethnicities, too. We have met representatives of all the other groups to discuss the matter, but it has not been an issue for them. It has come from the groups themselves: some deem it to be their ethnicity as well as their religion; others deem it to be their ethnic background but not their religion.

I represent around three quarters of Scotland’s Jewish community, and their representative groups have engaged with me on the subject quite a lot. Just to be clear, I note that there are essentially three broad definitions of what it means to be Jewish: there is a religious identity, a cultural identity and a recognised ethnicity. A person can be ethnically Jewish without being a practising religious Jew, for example, which is different from other faiths. I am a Christian and a member of the Church of Scotland, but that is clearly not an ethnicity; it is purely a religious question. I can answer question 21, which is on religion, and can identify as white Scottish under question 23.

The Jewish Leadership Council is very happy with where we have come to in the census. “Jewish” is included—correctly—as a faith, and then in question 23 there is a prompt for people who are ethnically Jewish to identify as such in the census. We know that one of the ways in which antisemitism manifests itself is in discrimination against people who appear to be Jewish on the basis of their ethnicity. Regardless of whether they practise as a Jew by faith, they suffer from antisemitism simply because they look Jewish. No position is ever going to be unanimous, but, on the whole, the Jewish community is happy with what we have in the census, and I am grateful to you for that.

However, I want to pick up on what you said about the Sikh community. I am not sure whether you are aware of this, but we have received correspondence on the census from Sikhs in Scotland. Sikhs are similar in this regard in that Sikhism is a religion but there is also a legally recognised Sikh ethnic identity. Given that the census quite rightly prompts people under section F of question 23 to state “Jewish” as an ethnic identity, why is “Sikh” not mentioned?

You will be aware that, for a variety of historical and cultural reasons, many Sikhs are not going to feel able to tick a box under “Asian” in section C of question 23, because they do not identify with that term. It appears that there is an appetite in the community for the inclusion of a prompt for Sikhs in section F. Will you expand on what you said about your engagement with Sikhs and how it relates to the correspondence that we have received?

Jill Morton, who has been involved in the issue, will comment on that.

The Sikh community approached us during the development process, asking us to consider whether it could have a separate tick box under the ethnic group question as well as under the religion question. Such a box is already included under the religion question, and we have no reason to change that. We have spent some time understanding the data needs and who the data users are for data on the Sikh community, and there is an identifiable need for data.

We spent some time doing a range of testing with the Sikh community in order to understand how people would respond if we had “Sikh” under the religion question and an additional tick box under the ethnic group question, and the result was pretty mixed. Although some members of the community were keen on that and could identify as Sikh under both questions, others found it confusing. In particular, we noticed in the testing that there was a difference according to age. We were consistently asked, “If you are going to have ‘Sikh’ under the ethnic group question, where are all the other Indian religions?”

We tested a variety of formats for where we might include a “Sikh” tick box for 2021, because not everybody would necessarily identify under “Asian”. In the various versions of the testing, however, we found that good quality data is collected on the Sikh population under the religion question. That is slightly different from the position when we tested for the Jewish community, because having only the religion question undercounted the size of the Jewish population. Census 2021 can continue to deliver good quality data on the Sikh community in Scotland through the religion question rather than Sikhs needing to be identifiable under more than one question.

I appreciate that. That is a useful answer. There is, however, clearly a need for some further engagement with the community. It would be useful if you could engage with the community and keep the committee copied in on that. There is still something to be resolved.

Kenneth Gibson made a point about “Muslim, please write in below” being an option for 2021. At the private meeting that you had with the committee, which enabled us to get some informal feedback, I raised that issue with you, and we talked about the fact that, particularly for some older members of the Muslim community, for whom English is not a first language, that is not clear. It is useful to know that a different wording will be used in the online version. However, I am still concerned that, in the paper copy—of course, I understand that, all being well, paper returns will be a small minority this time around—it is still not clear how a Muslim should answer that question. If you say, “Other Christian, please write in below”, in order to determine whether someone is from another Christian group, it is broadly clear what you are asking for. However, I am concerned that, if you say, “Muslim, please write in below”, people will not write “Sunni”, “Shi’ite” and so on. Rather, they are going to write “Muslim”.

Obviously, we can follow that up. If we were to use the word, “denomination” within that question, do you feel that that would be helpful?

I think that it would be, but I would definitely consult the communities. I use “denomination” because that is the language that I am familiar with, as a Christian who does ecumenical work, but I would not want to speak on behalf of the Muslim community. I assume that “denomination” is an appropriate word choice, but I would want to engage with the various Muslim communities on that point. A small change in the paper version will sort out the problem, and the online version sounds absolutely fine.

We will certainly do that and will follow that up with the communities.

What dialogue have you had with the Royal National Institute of Blind People Scotland, Visibility and other organisations that deal with sight loss?

That is definitely something that we have plans for. Just to clarify, are you asking about digital forms and paper forms, or only one version?

Both.

We have plans for accessibility testing of the paper form, and there has already been some accessibility testing of the digital platform. Some positives came out of that, as well as some things that we need to go away and consider.

There is accessibility testing going on across both platforms in relation to people who face barriers to completing forms.

Are you also consulting groups such as Dyslexia Scotland?

We have engaged with a number of organisations. I might need to get back to you on that precise question.

We are happy to write to you with details of whom we have engaged with. I know that the testing involved visual accessibility as well as language use, in order to ensure that the questions are clear. We got some good feedback on that. We are seeking to learn more about those issues from the rehearsal, because we want the process to be open and accessible to all.

I chair the cross-party group in the Scottish Parliament on visual impairment, and I am the deputy convener of the cross-party group in the Scottish Parliament on dyslexia, which is why I raised those issues. I look forward to your written reply.

I know that census day is 21 March, but for how long will the census be live online, and how long will people get to fill out the census?

The timescale is slightly different—shorter—for the rehearsal, but, in 2021, there will be a period of about nine weeks. The 21 March date is the reference date for the census, but we will contact people about three weeks before, and we will accept—indeed, encourage—early responses. We want to get people to respond as early as possible, but we also want to ensure that not everyone is responding at the same time. The census will remain open for six weeks afterwards, while we do follow-up activity and send out further communications encouraging people to respond. Gradually, towards the end of that six-week period, we will get into non-compliance activity for people who have not complied with their legal obligation.

After six weeks, we will start activity in the field on the census coverage survey, which is a separate survey to estimate the size of the population who have not completed the census. In total, there will be nine weeks of follow-up activity in 2021.

Did you learn to take that approach as a result of what happened in Australia in 2016, when its census failed?

Absolutely. However, the Australians’ model is slightly different from ours—they encourage people to fill in their census on the day. We have always had a reference date, which helps to balance things. Even in 2011, we encouraged people to fill in their census before that date. However, particularly with going online in a major way this time, it is essential to say to people that, although they are filling in the information as it relates to 21 March, that does not need to be done on 21 March.

In 2021, a lot of other things will be taking place in society. This Parliament will be dissolved and there will be an election. There is also the matter of what may or may not be happening with Brexit and its implications. What promotional activity will you be doing? What advertising campaign will you be undertaking to get over to people the message of how important it is that they fill in the census and do not forget to take part?

We are working on that with creative partners as well as Government marketing and comms colleagues. At the moment, we are doing a lot of research with members of the public to understand how they view the census and what messages will resonate with them. Ultimately, we want people to complete the census early, we want as many people as possible to complete it online and we want people to complete it without someone having to go round and chase them up. This is about people understanding the messages and our making sure that they are clear about the value of the census and what it is used for. That work will be going on over the next year or so, so that we get to the point of having an agreed campaign.

One reason for choosing 21 March, which is the earliest date on which a census has taken place, is to avoid the pre-election period for the Scottish parliamentary elections. It is really important that we get the message out to people nationally that they should complete the census before their attention is, understandably, placed elsewhere.

Before the census starts, we will enter a pre-census period in which the message will be that the census is coming. That will be about raising awareness. We will also work with community groups and others so that, when people get their letter asking them to complete the census, that will not be the first time that they have heard about it.

Alexander Stewart asked about the online technicalities of the census. I presume that you have financial estimates of the cost of the additional information technology infrastructure that you will require. For the 2011 census, you had a target of 20 per cent of people filling in the form online; for the 2021 census, you want to increase that figure vastly. Obviously, that will come at a cost. I also imagine that the planning for that will have to happen sooner rather than later.

Absolutely. Much of the infrastructure for the system that we are using is in place, and we will be using that for the census rehearsal. The online collection tool will be cloud based and secure, and it will allow us to respond to the demands of hundreds of thousands of people wanting to complete the census at the same time. That ability was tested in the lead-up to the rehearsal, but we will learn valuable lessons when using it in the rehearsal.

I want to pick up on the excellent innovation that it will not be solely the head of the household who completes the 2021 census. Everybody was very pleased to see that welcome development. In your rehearsal, how will you ensure that that works? What are your arrangements for ensuring good take-up of the flexibility, particularly among those aged 16-plus?

We allowed people to complete the previous census individually, but there was not such a big take-up, particularly online. Again, we are being very clear that people can, if they wish, request to give an individual response. The message that we are putting out for people who want to fill in their own census form is for those who have the agreement of the person who will fill in the main census form to do so—so that they will not be put on that form—and for those who do not wish to tell the person who will complete the main form. Arrangements are in place for people to do that confidentially, which will be signposted clearly on the website.

That is good to hear. For general awareness, do you intend to work with Young Scot and the Scottish Youth Parliament, for example, to ensure that young people know that they have those options?

We are working with a range of stakeholders for the rehearsal, and it will be important that we pick up on awareness raising, which we will do in the lead-up to 2021. We are keen to learn from the rehearsal what proportion of the population will take up that option, because it matters for how we will support them for 2021 and also from the point of view of processing volumes. We will continue to engage with partners who can help us with that.

Young carers might have an important role as well.

That point was brought to our attention recently, so we will make sure that we do work on that.

That is good.

We will make sure that people are able to provide that information in a confidential way.

I have a couple of questions about other issues. The census involves a combination of questions. Some are factual—when it asks about qualifications, someone either has 5 O grades or they do not—and some are based on self-identification or interpretation. Going back to the ethnicity question, the guidance says that it is up to a person how they interpret and answer it. It is interesting that the Scots language question, which was introduced in 2011, is a self-identification question, with the guidance listing “Borders”, “Doric”, “Fife” and “Shetland”. Members have been asking questions about accuracy, but this question is not about measuring accuracy; it is about what people feel is their knowledge of Scots and how they interpret “Scots”. How did that question work the last time, in 2011?

You are right to say that there is a range of questions, some of which are more factual while others, such as the Scots question that you have mentioned, are about how people identify. For 2011, a lot of testing was done of that question, because previous work on Scots language census questions had suggested that it is difficult to get to that level of granularity with one question. It is fair to say that the question that you have outlined is a broad one because people’s understanding of what is meant by “Scots language” is quite broad. Jill Morton may want to say more.

For 2021, we are continuing to work with stakeholders to refine the guidance that supports a range of questions, including the Scots language one, to make sure that we gather the data that meets their needs, as far as we can. As Amy Wilson has acknowledged, the census is a self-completion exercise, so we rely on everybody answering in the best and most honest way that they can.

Question 18 is about health, and it lists mental health conditions. The guidance includes “Addiction” on the list. Are you confident that that option will collect the information? I know that people can write in the “Other condition” section, but the question of whether that was the right place for that condition stuck out to me. Do people who fill in the census recognise addiction as a mental health condition? Has there been any testing of that, or has it been used previously?

We have not specifically tested whether people recognise addiction as a mental health situation. This is another example of where we are working with a range of organisations to refine the lists in the guidance to make sure that we have things in the right place. We can take your point away to consider.

Thank you.

I thank all our witnesses for coming in today. That is the end of our session.