Culture, Tourism, Europe and External Affairs Committee

Meeting date: Thursday, January 30, 2020

Official Report 643KB pdf

---

Subordinate Legislation

---

Census (Scotland) Order 2020 [Draft]

Good morning and welcome to the fourth meeting in 2020 of the Culture, Tourism, Europe and External Affairs Committee. I remind committee members and members of the public to turn off mobile phones. Anyone who is using an electronic device to access committee papers should ensure that it is turned to silent. We have received apologies from Mike Rumbles.

Our first agenda item is evidence taking on the draft Census (Scotland) Order 2020. I welcome to the meeting Fiona Hyslop, Cabinet Secretary for Culture, Tourism and External Affairs; Scott Matheson, senior principal legal officer for the Scottish Government; and, from the National Records of Scotland, Pete Whitehouse, director of statistical services, Scott McEwen, head of policy and legislation, Scotland’s census 2021, and Jill Morton, senior business lead, questions and collection instruments.

Members will be aware that the draft Census (Scotland) Order 2020 was laid in the Scottish Parliament on 23 January. A copy of the instrument is provided in the committee’s meeting papers. The committee’s consideration of the draft census order will follow the parliamentary process that is used for an affirmative instrument.

Today’s session provides an opportunity for the committee to take evidence from the cabinet secretary on the draft order. The committee will have an opportunity to vote on the draft order at a later meeting.

I invite the cabinet secretary to make a short opening statement of no longer than three or four minutes.

Thank you, convener.

I am appearing before the committee because, as the convener has stated, the Census (Scotland) Order 2020 has now formally been laid in draft before Parliament. We are therefore at a critical time for Scotland’s 2021 census. It is of significant importance that the committee gives the census order due consideration before it progresses to the Parliament chamber for approval.

For previous censuses, the laying of the formal draft order would have been the first point in the process at which the committee had sight of the proposed order. However, for the census 2021 legislation, the National Records of Scotland provided the committee with early sight of the draft census order in September of last year as part of an informal engagement process with the committee.

That engagement has been on-going for more than six months, since June of last year. The committee took evidence from officials in September and earlier this month. Members have also had access to a number of documents to support that early consideration, including drafts of the proposed questions for 2021 and all the impact assessments.

A demonstration of the online collection platform was also provided in September. In October, the National Records of Scotland provided written responses to all the questions that members asked at the demonstration session, and its representatives attended the committee’s evidence session on 9 January. My letter earlier this week provided further information that was requested by the committee following that session.

The new approach follows parliamentary committee recommendations from the previous census to improve the process and to enable a pre-scrutiny element for the respective committee. That was to allow any issues to be resolved before the formal process begins, thereby reducing the risk of problems and delays during the formal period. That is why the National Records of Scotland has been keen to engage with the committee over recent months and to provide information to support its considerations.

We need to have all the census legislation in force before this year’s summer recess, and the current process would allow for that. The first timing pressure for the legislation is the requirement to have questions agreed so that the online collection platform can be built for 2021. We must ensure that legislation is passed so that the timetable can be maintained, and a failure to do so would put at risk the delivery of the census in March 2021. Given that Scotland’s 2021 census is digital first, being predominantly online, that is a critical part of the preparations. Therefore, the regulations that follow the order need to be in force, as they will set out the format of those questions at the level of detail necessary for building the online platform.

I want to briefly highlight why it is important to support Scotland’s census, which will be held on Sunday 21 March 2021—subject, of course, to the order receiving the Scottish Parliament’s approval. For more than 200 years, Scotland has relied on the information that the census gives us. It provides vital information that Government, councils, the national health service and other users need. The information that is gathered from the census helps us to understand who we are, how we live and work, how we change and grow and the services that we need.

The key quality aspects of census data are that it has to be able to count the whole population, it has to be credible, people have to have confidence in it and it must be consistent over time and comparable across the UK so that it can inform collection and census outputs.

I am proud of the richness of the data that is held and the consistency of approach that can be demonstrated over those 200 years. That is why the National Records of Scotland has worked hard to ensure that the questions that are proposed for 2021 will provide accurate and reliable outputs.

I hope that, in the next few weeks, we can reach agreement on approving the census order. I think that we all recognise the importance of the census in providing information about a complex and changing Scotland. It is in our hands, collectively, to ensure that the 2021 census delivers for Scotland.

Thank you very much, cabinet secretary.

As you are well aware, the committee has spent some time considering the sex question in the census and the guidance that was introduced in 2011, without consultation, which suggests that the question should be answered on the basis of how one feels as opposed to what one’s birth sex is, and that a gender recognition certificate is not required if how one feels is different from one’s birth sex. That has caused quite a big national debate.

Last year, Professor Alice Sullivan from the centre for longitudinal studies at University College London, who is the director of the British cohort study, sent a letter to the Prime Minister, the First Minister and the census authorities across the United Kingdom, which was signed by 80 prominent social science academics who use population data. The letter raised serious concerns about the guidance. It says:

“The guidance acts to conflate two distinct characteristics—sex and gender reassignment, both protected categories under the Equality Act 2010—and will effectively transform the longstanding sex question into a question about gender identity. We are concerned that this will actively undermine data reliability on a key demographic variable, and damage our ability to both capture and remedy sex-based discrimination”,

which involves issues such as the

“different health and socio-economic outcomes between men and women”.

The letter goes on to say that the signatories welcome the new voluntary question on gender identity, which the committee unanimously approved of, and says that that means

“that there is no justification for advising respondents to give inaccurate answers to the sex question.”

Further, you will be aware that Professor Susan McVie, who advises the Scottish Government on statistics, told this committee that the 2011 guidance, which was introduced without consultation, was a mistake. Given that we are talking about prominent academics, it is strange that their views do not seem to be being taken into account.

I know that you understand what I am about to say, but, for the benefit of those watching this or reading the Official Report, it is important to state that the guidance does not form part of the order that has been formally laid; it is separate.

Of course, a lot of attention has been paid to the guidance. However, the census uses, and has always used, a self-completion approach. That means that people will complete it in the way that they themselves think that they should and in accordance with how they want to present themselves.

We know that the vast majority of people will not even look at the guidance. The NRS gathered important data on that from the rehearsal—during which I think that there were only 50 or so references to the guidance—and also from the subsequent research that took place. We know that 99.5 per cent of people will simply pass the guidance by and fill in the form on a self-completion basis, based on their biological or legal sex. We also know, from the early cognitive testing, that people have a different understanding of sex when it is discussed with them as opposed to when it is not discussed with them, and that most people will just think about it in terms of what is on their birth certificate or what their biology is. Most people will just answer the question quickly as they go through the form.

When it comes to the 0.5 per cent of people who refer to the guidance, a need was identified with regard to those who are not clear about how they would want to answer that question. That could be for a number of reasons. For instance, they could be non-binary—we know that those who identify as non-binary have concerns about how they might answer a question about whether they are male or female. There are also issues about people who are transgender and people who are just not sure. The point is, however, that the guidance is for 0.5 per cent of people.

I think that society is changing. The academics have concerns about how they can interpret data in terms of predicting where a country will be in the future and whether any trends can be identified. Obviously, those are future considerations, and what we need to do is capture information about the present.

I understand that there was a meeting between the NRS and three of the Scottish professors, including Susan McVie, at which clear concerns were expressed about the guidance. I am not sure that the academics have indicated what their suggested replacement would be. However, on the issue of whether there should be guidance, my understanding is that the academics would prefer to know the basis of the guidance than to have no guidance at all.

With regard to the letter that was sent to the Prime Minister and others—because it is not just the NRS that is producing a census; Northern Ireland is producing one, and the Office for National Statistics is producing one for the rest of the UK—I understand that the professor you mentioned and the various organisations involved had a meeting last week, although I have not seen the minutes of it.

I understand that people are looking at the question and wondering whether it will use a broadbrush definition. However, the census does not define what sex is, and it never has done. That question has always been answered on a self-completion basis. The census has never said that sex is biological, legal or whatever—it has only asked someone to say whether they are male or female. That is the process that we intend to continue.

There was an issue about whether, with this census, there should be an option for people to say that they are non-binary. Obviously, that question was debated fully with the committee, and there were debates on it in Parliament. Following that, I wrote to the committee to confirm that we would maintain the position of having a male/female question, as recommended by the committee.

That is the position that we are in now. Obviously, the engagement with the academics has to be serious. As I said, the guidance is separate from the order. Particularly at this point, because of the timings for the build of the online questionnaire, I want to concentrate on ensuring that the order is accurate and correct and reflects what people have said, and there are changes within that.

I understand that wider debates about sex and gender are taking place in other places for other reasons, but that is not the role of the census. The census is designed to capture information, and to be quite specific about doing so. The vast majority of people are interested in answering the question, which they do and do well. However, we understand that 0.5 per cent of people will need to look at guidance, and that is why the guidance is there.

09:15  

I think that the academics who wrote to the statistics authorities were well aware of the guidance and how many people looked at it, but they were still concerned about the outcomes.

In your letter to the committee, you respond to my question about which independent academics and other data users who use census data to support their research have expressly stated their support for a self-identified sex question. You mention public authorities, although it is not clear that they actually requested a self-identified sex question. However, on the issue of independent academics, you refer to a group of 50 who wrote to the committee on 20 September 2019, and you also refer to the Equality Network, LGBT Youth Scotland and Stonewall Scotland, which are not independent academic research organisations.

I would like to compare the letter that was led by Professor Alice Sullivan with the letter from the 50 academics you mention. The letter from Professor Alice Sullivan was signed by 80 senior academics—mainly social scientists, and mainly professors—who work with population data. The signatories include professors of research methodology, professors of public health, professors of medical sociology and professors of demography and statistics. However, although the signatories of the letter that you highlight include some professors—and I am sure that a small number of them use population data—it also includes academics from departments of creative writing, Atlantic studies, theology, computer science and linguistics. The signatories do not consist of population-data users; they are activists. They are entitled to their views, and I am sure that they are accomplished in their fields, but they do not compare to people whose specialism is in using population data. It seems strange that you have ignored one group of experts and preferred another group of activists when making your decision.

I think that that is an opinion. I am not attaching importance or otherwise to various groups. As a Government minister, I listen to all those who put forward views, not least very senior academics. That is why I think that it was important that, as recently as last week—I can confirm the date later—there was a meeting between Professor Alice Sullivan and the relevant statistical authorities.

My response to the committee about asking a binary question but having guidance for the 0.5 per cent of people who are looking for guidance that says that the question can be answered on a self-identification basis was issued prior to that meeting. You are suggesting that I was influenced by one group of academics whom you describe as activists. However, what I did was consider the research that was conducted with regard to what would ensure the best completion rate of the census and deliver the greatest accuracy. That research was shared with the committee.

But the research does not show that that approach leads to the best completion rate. The research, which involved a general population sample of 2,000-plus people, indicates that 3 per cent of people might not complete the census if there is a self-identification question in it.

You are extrapolating in concluding that that is because of self-identification. There are lots of reasons why people might not complete the census. For example, there are people who—

No—that is what the research says. There is a percentage of people for whom such a question might cause them not to complete the census.

That might be your interpretation. I will bring in Pete Whitehouse.

The basis is that the sex question, in past censuses and in testing, is very well answered. In 2011, it was the second highest answered question. There is no issue about how people will respond to the census and their willingness to take part in it.

When one does a particularly focused piece of testing, in an isolated context, whereby people are shown a certain question and asked for their response, one gets lots of different reactions. They might not be the same reactions as come through when there is a census that has a legislative background, in which there is a requirement to engage, and in relation to which the NRS has done all the compelling work that we will do over the coming months to show the benefit of the census, how it counts and its importance in enabling people across Scotland to be identified and to contribute. The evidence is that, in that context, people respond to the sex question.

The other thing to point out is that it is mandatory—

I am sorry, but what Pete Whitehouse said is not what the report by ScotCen says.

I think that we have the report to hand, so we can to quote from it; let me just find the information.

Maybe we can come back to that.

I will ask a final question. You talked about society changing. I certainly do not think that people back in 1921 or 1931 thought that they were self-identifying their sex when they answered the question.

I understood that Professor Sullivan did not get a response from the NRS; perhaps you can tell me the exact date when she met you. The academics, led by Professor Sullivan, warned in their letter of “extreme consequences” of subgroups, as opposed to people in the general population, self-identifying their sex, in terms of how that could affect our understanding of society. They pointed out that when it comes to certain subgroups, we are not talking about a tiny number of people. The letters says that, for example,

“1 in 50 male prisoners in England and Wales identify as transgender”

and that

“The Tavistock and Portman NHS Trust claims that between 1.2% and 2.7% of children and young people are ‘gender-diverse’.”

Professor Sullivan and her colleagues are saying that the approach could have a significant effect on data when it comes to subsets of the population.

I will ask Pete Whitehouse to respond to the question about Professor Sullivan.

On the technical point about the response, as the cabinet secretary said, the letter went to the offices of a number of individuals—the First Minister, the Prime Minister and the three registrar generals—as well as to the Office for Statistics Regulation. The response needs to cover all those interests, so there is a bit of process required to do that. There was a meeting—

You admit that Professor Sullivan did not get a response.

I am just explaining where we are.

There has been a meeting between Iain Bell, the deputy national statistician, and Professor Alice Sullivan, which the NRS phoned into. That happened last week. Professor Sullivan was aware that the meeting was happening and was happy to engage. Following that, the response is being drafted to take account of those discussions, as well.

Dialogue is happening. What we have is a letter that went to five or six different offices in different places across the UK, and we have to talk to one another in order to respond. Engagement absolutely is continuing: that meeting with the deputy national statistician happened.

Is he deputy national statistician at UK level or in Scotland?

He is the UK—

So that meeting was arranged at UK level; Alice Sullivan did not get a direct response from anyone in Scotland.

But the NRS was part of that meeting.

The NRS was part of the meeting—it was on the call. That happens frequently with UK offices. If someone sends a letter to the UK Prime Minister and Scotland’s First Minister, a co-ordinated response from the two does not happen immediately, as you will realise; it involves lots of different people. I do not think that we should make a judgment in that regard.

Okay—but Professor Sullivan did not have a meeting with anyone in Scotland.

Before I move on to a supplementary from Kenneth Gibson, I point out that the meeting that you had with a couple of the signatories, which you mentioned earlier, was before Christmas. That was with Professor McVie and two others. You produced a minute of that meeting that the academics were not happy with. We have documentation showing that there is not a lot of agreement between yourselves and those academics.

I have not met them, although the NRS has. On the minutes, I have questioned whether we have a good understanding of what the various positions were. One of the questions was whether there is guidance or not. I was not party to that meeting. I might ask Pete Whitehouse to say whether the minutes have now been agreed with those individuals.

The meeting happened, which was very useful and helpful to us. As happens in lots of other organisations, the NRS—

I am sorry to stop you. You say that the meeting was “useful and helpful” to you, but what has it made you change? If the meeting was useful and helpful, what have you changed as a result of it?

There was clarification in the meeting about the importance of guidance, which is a valuable asset to people who look to use analysis. There was a conversation with Professor Nick Bailey, I think, on some areas where guidance across the census might have been inconsistent, so we have tightened up some of the language. There were some broader discussions, which probably reached into areas beyond the census, concerning wider issues about what is happening in society, as the cabinet secretary has just mentioned, and how one deals with these sensitive—

Would you say that the academics are happy about the meeting? They have disputed your own minutes of it.

Sorry. I was going to go on to say that the NRS took a note that it provided to the other participants, as is normal and as is our practice, and they then said that they did not agree with some points. In order to be absolutely open on all of this, we provided the committee with that full narrative, so the papers that we sent you—a while ago, I think—included our note, the comments that came back from Nick Bailey, the comments that came back from Susan McVie and our responses to those. All of that is packaged and provided to you. It is not the case that we are saying “They’re wrong; we’re right,” and so on. We are giving the committee the full record of that engagement, which I think is an honest approach.

I think it probably shows that they are still not satisfied.

You said earlier, cabinet secretary, that the census does not define what sex is. The proposal is that, for question 3 of the census, we will have “Female” and “Male”. The current proposed guidance, which is being tested, says:

“If you are transgender the answer you give can be different from what is on your birth certificate. You don’t need a Gender Recognition Certificate”.

You have also said this morning that it is important to capture accurate data. How does it help to capture accurate data when folk are effectively encouraged to put whatever they wish for that answer, given the conflation of sex and gender there? The committee has debated that point long and hard, and the view of the committee is that sex and gender are not the same. Yet the guidance seems to indicate that, for the question as it is currently being proposed, the two are almost the same.

For the vast majority of the population, it is not an issue. We know that 99.5 per cent of the population will complete that question on a self-completion basis without referring to the guidance.

The issue about the census is that it is a civic responsibility—everybody has to complete it; it is a requirement that people complete it. That is why the census order is here before you. Certain questions are mandatory, including the sex question.

In relation to the 0.5 per cent of people who need guidance as to how they complete that question, I point out that the census has always operated on a self-completion basis. We cannot go round to everybody’s house to check whether people have filled in the questions accurately. You are asking about accuracy. Accuracy is not policed on an individual basis—we cannot do that. For us to help people to fill in the census, they have to know that it is confidential, that they can fill it in themselves and that they can do so honestly, from their own perspective. It is a civic responsibility: we require people in society to answer the questions honestly.

We know from the testing that was done most recently that 25 per cent of the people who access the guidance really need to understand what they can and cannot do from their own perspective.

09:30  

Yes, but given the fact that sex and gender are not the same, surely the alternative proposed guidance that has been tested, and which is much clearer, is better. It says:

“The answer you provide should be the same as your birth certificate. If you have a Gender Recognition Certificate (GRC) you may record your recognised legal sex.”

The following question in the census is the trans question, so I do not understand the purpose of having guidance that effectively says to people, “It really doesn’t matter what’s on your birth certificate. You can answer any way you like.” How is that going to provide the accurate data that, according to the academics who the convener has quoted, is essential?

You have mentioned the figure of 0.5 per cent a number of times as if it is insignificant, but that is 27,000 people in the Scottish population. Even if the percentage is as low as 0.5, that is still a significant number of people.

Yes, and that is why we want to capture them. That is the issue. Discussions that I recall from before 2011 were about people who did not want to or could not fill in the questionnaire because there was not a question that transgender people could answer, and they did not have an understanding of how they could answer the question that was there. That is why the guidance was put forward in 2011 by the ONS, although, as the convener said, it was not tested at the time.

The point is about improving the quantity of those who complete and the quality of the data. If people get to the sex question and say, “That’s it, I can’t answer this; I’ve got nowhere to go”, they will not complete the rest of the census. Of course, it is mandatory and we want people to complete it—we want to maximise the impact. You are right to say that the figure of 27,000 people is not insignificant. Should we include them or should we say that we are not going to do anything to give them help with that?

The other thing about the guidance is, because it is going to be online, we know that more people might require help with it. There will be phone lines. I visited the phone line operators during the rehearsal, and a very small number of people phoned up about the guidance in that area. I do not have the figure to hand, but I think that I quoted it earlier as about 50.

Do we want to include those 27,000 people? Will having self-identification guidance for them help them complete the census? We know that the answer is yes.

We were also asked to test the option of having the guidance ask about legal sex. The research on what would maximise the census returns, which we shared with the committee in early January, shows that guidance on self-identification of sex would have more of a return than guidance on legal sex.

I do not agree that there is nowhere for transgender people to go in the census, because the next question is about whether people consider themselves to have a transgender identity. Reading the guidance that has been proposed, it appears to me that the Scottish Government does not believe that there is a difference between sex and gender. That is the clear implication of that question.

I do not agree with that, and that was not the case in 2011, either. One of the issues around the mandatory element of the sex question was that there was nowhere to go. If the choices had been “Male”, “Female” and “Other”, there would have been somewhere else to go.

If, on a mandatory question that is online, the only answers are “Male” and “Female”, and you do not complete that question, you cannot get to the transgender question. We want people to get to the transgender question precisely because of the concerns that we have. We do not know how many transgender people there are, and we want to get that information. The nature of the census means that you must be able to answer the male/female question before you get to the transgender question.

Lastly—because I know that others want to speak—why can someone not answer that question? Even if they feel themselves to be something other than the sex that they were born with, surely they can still say what is on their birth certificate. I am struggling to understand why some people cannot do that. The question is compulsory, so surely we would expect people to answer it.

The officials will correct me if I am wrong, but people could do that if they wanted to. That is precisely why the guidance says that there is a transgender question coming: so that people know that they will be able to indicate their transgender status in the next question.

Some people might choose to answer the first question with the sex that is on their birth certificate and to say that they are transgender when they answer the second question, but others might decide that, whatever is on their passport or whether or not they have a gender recognition certificate, they really want to put in the first question what their lived sex is. However, the transgender question can only be reached if you fill in the male/female question.

The case that has been put to the committee during the evidence sessions is that the sex question has always been a matter of self-identification. Some people would challenge the conflation of self-completion and self-identification. I am not sure that every committee member is convinced that there was confusion in people’s minds when they had to answer the sex question in 1911, 1951, 1961 or 1971. That question has always been a matter of self-completion, although I accept that that was not always checked, but the argument about it being self-identification is difficult to accept, as that description is a modern term.

There was no guidance on the question in 2001. The committee asked for the testing to include a situation in which there was no guidance. That was not done, although ScotCen collected evidence on who had not looked at the guidance. Why has the no-guidance option not been tested or considered? I understand that, in 2001, somebody queried a question and contacted the NRS for advice on how to answer, and the advice was emailed or was perhaps verbal. How was the decision taken on what advice would be given at that time? That decision created the sequence. In 2001, someone was advised that they could self-identify in their answer to the sex question and, in 2011, the guidance appeared, with no consultation. That is what has brought us to this stage.

I should just say that 2001 was definitely before my time. Every census has different issues and controversies: the issue in the previous one was around Scots language and, in 2001, there was a lot of focus on the ethnicity question. My recollection is that, when the census was going through in 2001, the need for advice was identified because of people querying questions. If you are one of the enumerators on the phone or going around, and somebody asks you for advice what do you say—nothing?

I agree with your initial point: we cannot say that self-completion is the same as self-identification. However, the basis for the census has always been that we trust what people say, and that they complete the form as they see it. It would be wrong to say that there were no trans members of society in 1921 or 1931.

I accept that there were people who were trans, but we cannot tell the underlying understanding of what sex meant at that time. We can look back at previous censuses, but we cannot tell—that is part of the NRS’s argument. I think that people understood and recognised that the sex-identification question meant their biological sex, whether or not they felt that they belonged to it.

That is what we are trying to do. Whether we are talking about the census in 1921, 1931 or 2021, how do you answer that question if you are trans? The vast majority of the population will self-complete—they will decide how to complete the form without looking at the guidance. We know from the testing that has been done that the vast majority of people do not look at the guidance. Is the fact that they do not look at it but still complete the form not a test of a situation in which there is no guidance?

When the initial work took place in August 2017, cognitive testing was done on questions that did not have guidance, and there were subsequent quantitative household tests. My NRS colleagues might be able to help with that.

Cognitive testing, whereby people were presented with a question with absolutely no supporting material, was one of the early starting places for the development of the questions. We found—similarly to what we found all the way through—that the vast majority of the population do not query a question on sex when they see it. We take that as a standard answer. However, the concept is not necessarily straightforward for a small number of the population, and they ask us “What do you mean by this question? What are you asking me? Is there guidance I can look at?” That was our starting point.

In the large-scale quantitative testing of a number of questions that we undertook in 2017-18, question guidance was not provided, so there has been testing of the questions without guidance. The most recent testing, which was published at the end of last year, specifically tried to replicate census conditions and environments. People received a survey to complete, largely unsupported, in their home. That was done to try to understand whether people use the guidance, who uses the guidance and the impact of different versions of the guidance. That is what the most recent testing achieved.

The convener raised the issue of non-response. I think that she referred to statistics relating to the question being tested with males and females, who were then shown the guidance to see what impact that would have on the response rates.

I do not know whether you have a copy of the ScotCen Social Research report in front of you, but I am looking at table 3.16. When the self-identification guidance was presented along with the question, 4 per cent of females and 7 per cent of males thought that the guidance was unacceptable, and there was a non-response rate of 2 per cent. There were similar percentages when people were presented with legal sex guidance: 5 per cent of females and 6 per cent of males thought that the guidance was unacceptable, and there was a non-response rate of 2 per cent. Do you have concerns about those figures and the non-response rate? Are you concerned that the guidance might make it more difficult for people to answer the questions?

I know that there is an argument that the majority of people do not look at the guidance, but it is a digital-first survey—the convener and I saw the trial run of the digital version—so people can see that guidance is available. People might, out of interest, click on the guidance. Will that influence how confident they feel in responding to the questions?

That goes back to the point that the testing was done to find out people’s responses in certain circumstances, including when they saw the guidance. We know that in rehearsals of census completion, only 50 out of 72,000 looked at the guidance. The 3 per cent figure relates to when people are presented proactively with the guidance and give their views. That is not what happens in the actual census.

The law requires that answering the question is mandatory, so people who are filling out the census will not be able to progress without doing so. I do not think that there can be a read-across between that research and the rehearsal experience, which is a more accurate representation of how people would receive the guidance.

In practice, only a few people looked at the guidance. We are comparing two different things. In the ScotCen testing, everybody looked at all the guidance, but we know that, in the real-life situation, people will not necessarily look at it. There is a difference when people look at the guidance. The ScotCen report shows that people said, “I’m not going to continue,” or “I’d prefer to skip this question,” but there is no skipping in the census. That is my understanding of the report.

I want to go back to my initial question on the guidance and the situation in 2001. This question is more for the NRS. When somebody contacts you, either by phone or email, what is the process for agreeing what guidance or additional advice is given? I am thinking about someone who sees the census and all the published guidance, but still has a question and contacts the organisation. When someone contacted the organisation in 2001, they were given a response. Advice was given out in 2001, so I am not sure why it was decided that the guidance should be created and given as the response to initial queries.

That was 19 years ago. I am not sure that anyone is in a position to answer that.

09:45  

Is anybody from the organisation able to explain that?

Can the NRS describe what happens now?

What would be the current system?

I cannot talk about 2001, but in general people would either have phoned up a helpdesk or would have been given written guidance on questions. If people needed further advice they would probably have spoken to the field force, who were the people who handed out the forms—we feel that most people will receive the census in a totally different way this time—or they may have phoned a contact centre.

In 2021, people will be able to access the guidance online, speak to field force people in their area if they are filling in the form on paper, or phone a contact centre. We are trying to ensure that the advice given is consistent across all those platforms so that there is clarity on how people should answer that question.

I want to explore two issues, the first of which is around consistency. One of the most compelling arguments for the proposed guidance is consistency with the rest of the United Kingdom, given that next year there will also be censuses in England and Wales and in Northern Ireland. Are panel members aware of what the other devolved nations are doing with the sex question in the census? Have they liaised with them? This debate must be happening there too.

There has been regular contact. They check what we are doing and vice versa. One of the arguments is to have consistency for comparability. There has been more scrutiny and questioning and more additional testing in Scotland, probably because of the endeavours of this committee. We came to decisions independently, but professional statisticians such as NRS will want to know what the ONS is doing and vice versa.

I have kept an interest in what the other Administrations are doing. In England and Wales, the ONS is carrying out the same questions with the same guidance: a binary question, with self-identification, and guidance for the small number of people who want to access it. We understand that Northern Ireland is doing the same. It has been an iterative process: we have been checking what they are doing and they have been checking what we are doing.

Is there any scope for divergence among the nations?

Absolutely. If we wanted to diverge, we could, but we have taken a decision that is backed up by the research that we have recently conducted into what would maximise the quality and quantity of data being captured. It has made sense to be similar.

I have not had direct conversations with the ONS. Pete may want to say something about that relationship.

We have a very strong and close dialogue with the ONS and our colleagues in the Northern Ireland Statistics and Research Agency. The Office for Statistics Regulation, which oversees the branding of national statistics and covers the whole UK, sees UK comparability and consistency as hugely important. It hears a lot from users and interested parties on the ability to have that comparability and consistency, so we are mindful of enabling it.

We should add that we have looked at what other countries have done, particularly on digital first, because having a digital-first census is a massive step.

There is also a question of consistency with the 2011 census and what happened before that time, although I accept that Claire Baker and others have made arguments about what people will do when they fill in the census if there is no guidance.

We have spoken about testing the question, or the failure to do so. As a matter of principle, is having no guidance on the sex question a viable way forward?

It would not be viable. Those who refer to the guidance, in particular academics, want to know the basis on which a question is there and the basis on which the census has been completed. In addition, it would be unfair to users. The whole point is that we know that some users need the guidance, without which we would be saying that we will not service them and that although we know that they have a need, we will not help them to fill in the census. There is a responsibility to those people.

Do other members of the panel have views on that?

I echo the cabinet secretary’s point. We are providing guidance to enable all people across Scotland to engage with the census, and we have evidence that support is required. As with guidance on all other questions, the purpose of guidance on the sex question is to help people who need that support to fully answer it.

As the cabinet secretary has just said, data users feel that it is valuable to understand the basis on which a question has been asked. Those are two compelling reasons to enable full completion, access and engagement with the census.

Can you confirm that the letter from Professor Sullivan and the 80 academics did not express a view as to whether there should be guidance or not?

I was specifically referring to the conversation with three of the academics who are based in Scotland, which is contained in the papers that were provided to the committee. They made that point in our discussion.

As Government statisticians and people who work with academics, we have a general understanding of how we engage with data. Understanding, through information that is provided in metadata and elsewhere, the basis on which a question has been asked is a massively important part of how analysts do their job.

You say that those three academics mentioned their view in a conversation with you, but we do not know whether they would choose to have guidance on self-identification or no guidance at all.

Nobody is saying that we do.

Therefore, we are going back to three academics as opposed the 80 who wrote to us.

I do not want to undermine the reputation or status of the three Scottish academics. They are well respected.

There is a general principle that in any work in which an analyst looks at a data set, almost as a matter of course, they will want to understand the basis on which that evidence was collected. That applies to health, justice, education, survey and administrative data. They will want to understand what the question was, how it was asked and on what basis. That is what we are trying to support.

Good morning. On the issue of the dialogue with the independent academics, it has struck many members, over the consideration of our current position and the previous legislation, that there seems to have been a failure by the NRS to engage early doors with people of that ilk. Perhaps we would be in a different place now if that had happened. Maybe you would like to get Mr Whitehouse to respond to that question.

I am not sure what “people of that ilk” means.

We are talking about those independent academics who will be using the census data, rather than about the people who are august in their own fields, but who, as the convener suggested in her comments, are perhaps not using census data day and daily.

We can go back through previous Official Reports and our questioning of the NRS, but it seems as though the engagement with people who took a slightly different view came very late in the day. I wonder why that was. I asked Amy Wilson that question at a previous session, but I never got a clear answer.

There are two issues here: the general involvement of academics in preparation of the census and their involvement in the sex question and guidance on it.

Yes—the issue of how we have got to where we are now.

The NRS’s approach has been open and it is open. Our organisation publishes and produces the census. We have not shied away from engagement. Anyone who has wished to engage with us over the past decade or before that knows where we are. We are publicly available and we come to the committee. There have been discussions on the television and in newspapers and we have received lots of letters. People have engaged with the national statistician of the UK Statistics Authority and the chief statistician in Scotland. We have been very open. We have an office in Ladywell house up the road. We are accessible. We have also carried out all sorts of testing and invited all sorts of different groups and individuals to be part of that process. We have welcomed everybody’s contribution.

With respect, on behalf of the NRS, I have to say that I do not recognise the idea that we are not open and that we are not available to have discussions. We are an organisation with a long history of engaging with individuals and groups, however they choose to come to us. Certain groups engaged at the UK level—perhaps because there has been a misunderstanding about Scotland’s ability to run its own census. We use our links with NISRA or ONS to access those groups.

Individuals and groups will have come to the committee or to members as individuals, and you have forwarded that information to us, and we have engaged with that. We have also received letters and emails about and have done lots of testing and held public consultations on those issues. Unless there is other evidence that shows that we have not engaged—if there is, I would want to look at it—I feel that we have been incredibly open, as we are expected to be.

I hear what Mr Whitehouse has said. Obviously, I made the point of referring to engagement early doors and if Mr Whitehouse wants to go back and look at the Official Report of our committee deliberations over the piece, he will see exactly what I am referring to.

I will move on. Question 3 on page 28 of the guidance has two options: option A, which is the current proposed guidance, and option B, which is the alternative proposed guidance—

Sorry, but what you are referring to?

Sorry. I am referring to question 3, which is the mandatory sex question, on page 28 of “Census 2021: Paper Question Set and Guidance”, the document that the NRS supplied to us for this meeting. That has two options—I will call them options A and B.

First, I would like clarification. I think that we have been told that we will not see a final text until the autumn. Why is that? I presume that you need to have the text finalised so that you can get all this online—the cabinet secretary made a similar point about another matter. I assume that the guidance will be made available to online users, and I presume that the same imperative applies. Why will we not see the final text until September? Is it September?

I am giving evidence today on the draft order, which sets out the subjects to be included in the census. This is the first step in making sure that the content of the census is agreed. It is open to the committee to modify the content. What is in front of you now will be what is in front of you when you vote in a few weeks’ time. The parts that are in italics are what you can change.

Then there are the regulations. Those will include the census questions.

It is quite important to get the timing accurate, to make sure that everyone has a good sense of it. I understand that, subject to the committee’s timetable, you will consider the draft census order, which is the subject matter of the census, on Thursday 27 February. The idea would be for me to go to the Scottish Parliament in March, with the view that the census order would go to the Privy Council on 15 April. After that, will come the regulations—those will have to be agreed to within 40 days, as you will be familiar with—which will have the questions.

The guidance is completely separate from the legal processes that we are considering. There is always some flexibility and movement in finalising guidance. From an information technology point of view, it is the content of the actual bits for completion—the questions—that are imperative for the build, which is what I think that you are referring to. Officials may wish to add to this, but the guidance is the text that goes round that, which is less of an imperative from an IT point of view and a build point of view. The imperative absolutely must be our having the questions ready.

I am sorry that that was quite a long answer, but I thought that it would be helpful.

10:00  

I understand that you said in your letter to the committee of yesterday that the final version of the guidance would be

“agreed by the end of summer this year.”

The last time that your officials were before us, we had a discussion about seeing a copy of the latest draft text. As I said, an option A and an option B are given on page 28 of the document that your officials have provided. Option A is effectively self-ID; option B is not.

Regarding option A, to repeat a point that I made to your officials a couple of weeks ago, the Scottish Government has embarked on a consultation that would effectively move away from the current gender recognition certificate approach to self-ID. If that is the case, how can a Government document pre-empt the result of that process effectively by saying that, for that purpose, there will be a self-ID approach? That would be my concern. It could be argued that that would usurp the role of the Parliament. The document is a guidance document; nonetheless, it will be issued further to the Scottish Government. If we are having a consultation about moving away from the current approach to self-ID, how can we be inserting a self-ID approach in advance of that process?

The census is a completely different issue and process from any process about gender recognition. For the census, people do not have to provide their birth certificate when they fill in their date of birth and they do not have to provide a gender recognition certificate when they are filling in the sex question. They do not have to provide their birth certificate when they are answering “Male” or “Female” to the sex question; it is about self-completion. It is not policed. It is really important that the process is confidential. People are asked to provide a lot of sensitive information, and not just in these questions. We have spent a lot of time on these questions, but there are other questions, too. It is not a case of pre-empting anything. If that was the case, we would say that the 2011 census did that, but I do not think that the 2011 one was pre-empting—

We have heard that there was not any consultation on the 2011 text. That was the first time that such an approach had been adopted since the census began 200 years ago. There was no consultation then. That is why we are where we are now: we are having the consultation, which I think is a good thing. My point remains: whatever the approach of an individual is in responding to the mandatory question 3, the guidance is, nonetheless, a piece of paper issued further to the Scottish Government, and that means that a self-ID approach is being pursued through option A, even though we are having a consultation on self-ID that has not reached any definitive stage. We are pre-empting the role of the Parliament in that very important debate if we pursue option A. Of course, there is the alternative of option B, which reflects the current legal position, and which I think is a much more appropriate route to go down, because we have to live within the context of the law as it currently applies.

How, then, will things go in the months ahead? Who will make the decision between option A and option B, and on what basis?

The NRS will consider that question, and it will also come to me—I will have a decision to make on that, too. If you are saying that someone will require a gender recognition certificate to answer the sex question, by that logic, you are saying that someone could complete the transgender question only if they had a gender recognition certificate. I am not sure that that is what the committee’s view was—unless I missed it.

No—we are all absolutely supportive of the voluntary question on transgender and sexual orientation.

And do you think—

With respect, I think that the transgender question is entirely different. Kenny Gibson mentioned that this issue is about whether you conflate sex and gender identity or you do not. The voluntary question on transgender does not conflate sex and gender; it is on gender, whereas question 3 is on sex.

Those are the issues that I have raised, and I would imagine that you will want to reflect on all the issues that all members of the committee have raised. If we get to the stage where, for some reason, the Government picks option A, and we accept what your official has said and what Scott Matheson said the other week, that would have no legal import whatsoever for any wider debate on the issues. Would it not be appropriate to make that explicit in the text of the guidance?

Sorry? What do you want to make clear?

If the Government—or the NRS or whomever—picks option A, that would pre-empt a debate that we are apparently about to engage in. Doing so would jump the gun. If it is the case that this is to be viewed simply as guidance, it has no legal import, it does not matter and it is just for the purpose of making people feel happier when they are filling in a form, would it not be appropriate to have a sentence to that effect in the text of the guidance? Is that something that the NRS will consider?

Everything can be considered in relation to the guidance. It is a recommendation about how people can answer. We are not saying, “You must answer it. This is the law.” There is nothing in the guidance that says that that is a legal interpretation of what gender identification or self-identification is, or what the status of that is. That is not the purpose of the guidance. The purpose of the guidance is to try to maximise the number of people who complete the questionnaire.

If option A is picked, I do not see why it would not be appropriate to make that point as clearly as the cabinet secretary has just made it.

I definitely hear the points that you are making.

Okay; thank you.

Did you say in response to Annabelle Ewing that you still have to make a decision on the guidance?

We said that in the letters that we have given to you about the timescale. You have seen—

No, I am talking about the sex question. Are you saying that you have still to make a decision?

I set that out in my letter to you when we sent the results of the research. I said that we would have the binary question with, for those who need to access it, guidance—from the research that we have done, we think that that will be for the 0.5 per cent of the population who will answer on a self-identification basis. That is the same as ONS and it is the same for Northern Ireland. That is a proposal.

That is what you say in the policy impact report. You have already made the decision, so why can you not share the guidance with us?

But we have.

It is public. The published report of our testing that we provided to the committee last December sets out our proposed guidance for the sex question. It very clearly says that that is our

“Proposed sex question and online guidance for Scotland’s Census 2021.”

Right. So that is not going to change?

That is our proposed guidance.

That is our proposal.

I turn to the first question that we asked in the letter to the cabinet secretary. The convener has covered some of the points about the differing views that academics have about the sex question. I am interested in the second part of that question, about other data users, particularly organisational data users. Were any concerns raised by public sector bodies about the proposed approach to the sex question, or should I take from the answer in your letter that they were content with the proposed approach?

They were content.

I want to consider other organisations that make use of the data—charities, non-governmental organisations, businesses and so on. Could you give more detail about the engagement with organisational data users outside of the public sector, and about what their response was? Did they raise concerns or were they also content with that approach?

I will ask the NRS to answer that, but I reiterate that the NHS and local government are major users of data in the public sector—they make considerable use of the data.

Our work in showing why the census is so important includes talking about its benefits. As part of that, we received plenty of feedback from the 2011 census about how people are using the information from it in their communities to seek funding, identify issues and help with analysis across all the variables in the census. We have received good information about how various organisations, including charities and those in the third sector, have made use of census data.

There has been the opportunity for people to have written to us to express their views. I am not aware of too much detail on that. However, I have not received anything directly to suggest that third-sector groups or charities that have previously successfully used the census have major concerns about how we are asking the question. Perhaps Jill Morton can add to that.

Certainly. The process started with the topic consultation, which was widely advertised. The registrar general sent out a lot of letters, and the consultation was promoted to get a wide range of engagement. We got feedback, specifically on whether the 2011 data was useful, on whether it met people’s needs and on what would need to change. Following that, we had several years of stakeholder engagement with individual groups and collective groups of stakeholders. We held a range of public events, including in Edinburgh, Glasgow and Aberdeen, which were promoted in order to get as many people to attend as possible. At those public events, we put forward our proposals and said where we thought we were with question development. There has been a wide range of engagement.

That answer relates more generally to all the questions. I am not sure whether Ross Greer’s question is just about the sex question.

It is primarily about the sex question, but I am also interested in the overall context.

Reflecting on what Annabelle Ewing asked about, I think that, because question development starts years and years before the census takes place, that is where the topics and issues come from. That is when people engage—and people who have always had an interest in the census will always engage very early in the process. For example, the veterans question has been included following suggestions from veterans groups. People scope their interests at the start of the process.

That is a general answer about what happens across census development, but I think that you want to focus on the sex question.

Although there is a substantial debate on the sex question among academic data users, organisational data users are content and did not raise concerns. The cabinet secretary made the point that the most substantial and consequential use of the data is by public sector bodies, which are content with the proposed approach. Is that a fair characterisation of the situation?

Yes, that is fair. However, I would not want to dismiss the needs of researchers and academics.

Absolutely.

Some of the issues might be about how society is moving and changing. I reiterate that the census cannot predict the future; it must reflect the current situation. We cannot see trends unless we capture information on where things are now.

Absolutely. I accept that, before 2011, there was a de facto no-guidance situation. I realise that this is not what is being proposed and is a hypothetical example, but, if a conscious decision were to be taken to ask the sex question with no guidance, and if someone were to receive the census, be unsure how to answer and phone the NRS and ask, “How should I answer this question?”, what would happen?

I will ask Pete Whitehouse to answer that. It is a hypothetical example, because that is not what we are proposing.

That is a live question. To be honest, I do not know how one would respond to such a question. If, in that hypothetical situation, there was no guidance and someone sought guidance, we would be in a catch-22 situation. The NRS would have to do something.

The person would say, “I’m not allowed to say anything,” or “I can’t tell you—it’s up to you.” That goes back to Claire Baker’s point about whether it being up to the individual means that it is self-completion or self-identification. It will depend on how they interpret the question.

Yes. There will be a reputational issue for the census itself, for the NRS and for the Government if people are phoning up and asking for help but are unable to get clarity on how to answer a question.

Yes.

10:15  

On the matter of testing, among the general population group of 2,000 who formed part of the ScotCen testing, what was the division between those who filled in a paper copy of the census and those who used the digital version? Was it roughly 50:50?

I do not know that off the top of my head. We can come back to you on that.

I will explain why I am asking. I am interested in whether there were any notable or significant differences between the two groups—those who answered online and those who answered on paper. The division in testing interests me, as the intention is to have an overwhelming majority of people complete the census online.

It depends on what you are asking about, but the rehearsal might give you better information about how people responded. Some people could request paper copies. There is a differentiation in how people completed the census in that the rehearsal period was earlier than the autumn period. We know that 99.5 per cent of people who answer online do not look at the guidance. That is real, live testing, and the rehearsal probably provides the strongest evidence of how people behave with guidance.

If you do not have the information immediately to hand, I am sure that you can come back to us with it.

I have now located that information in table 3.2 of the ScotCen methodology report. For the unweighted bases, the paper and online responses were split roughly 50:50. There is no difference between the online response and the paper response in who identified as male or female—the split is about 50:50 as well.

On a methodological point, the intention is not that the census responses will be split roughly 50:50, but both groups were still large enough and sound enough samples to be statistically valid, and each one was informative.

Yes.

Absolutely. There is validity there. As I think I said previously, ScotCen carries out the work on our behalf, and it did that to the methodological standards. We are content that it meets the criteria.

Great. Thank you.

Alexander Stewart will move on to another area of the census. Do you also want to ask about the sex question, Alexander? It would be good to wrap that up before we move on to another area.

You have answered many questions this morning, cabinet secretary, but the clear point that we have got is that you want the most accurate information on all of this. However, what we have heard about the guidance, the lack of guidance and the possibility of different types of guidance has possibly confused the whole issue in many people’s minds.

When the testing was done—you said that there were lots of different types of testing—and there was the rehearsal, the guidance was much more prominent online. It is not that it leads people, but it is there—it is identified that guidance is available when people come to a question. Your letter says that

“The purpose of the testing NRS commissioned ... was to understand the manner in which respondents might make use of guidance”

in answering the sex question. What have you identified from all of that testing? What has been the main issue with the guidance that has been in place? Can you tell us of any real situation or scenario that represents the purpose of that work and that has shown what has been achieved by all the testing and rehearsing?

The report that was sent to the committee set out the responses, particularly with regard to the ScotCen research. The combination that would achieve the best response—that of more people filling in their answer and continuing—is the binary question with self-identification guidance.

We understand that that is the process that will take place, but there is still unease among everyone who has participated in the process, including the academics. The number of questions that we have had and the time that we have spent on the matter today indicate the difficulty that is anticipated if the guidance is not in place and does not cover everyone. It is apparent that not everyone is happy with the whole process. That is the problem that we face.

Clearly.

We have to unravel that. Even today, I am not confident that we are at the stage of unravelling that to make sure that we get the best and most accurate data.

We have to look at the information that we have and the research that we have done. We have done far more research than any other part of the UK to understand the issue. I am not dismissing the fact that there is a general debate about it, but there is a danger that people are conflating different debates. As the Government minister in charge of the census, my responsibility is to make sure that the census is delivered on time, that it has the maximum completion rates and that it can be used primarily by service users but also by those who complete it. That is my job. I am not here to lead debates in society—I said that clearly at the beginning of the debates that we had in Parliament. It is not the job of the census to lead society. The job of the census is to capture information about how society is now.

Thank you.

We have a couple of supplementary questions on that topic.

I want to clarify what Mr Whitehouse said about the two versions of the guidance that we have been given. The guidance says that

“This question is under development.”

However, you said that that is the proposed guidance. I am struggling to understand why we are still being presented with two versions, because you seem to be firm in saying that that is the guidance that you intend to use. Why are you bothering to give us two versions?

That is what was tested.

But it is presented to us as if there is still an on-going discussion and a decision is still to be made. If you have made a decision, it would be helpful for us to know—

Our recommendation report, which we provided following the testing that ScotCen did, set out our recommendation. At other times—and perhaps alongside that; I cannot quite remember—we also provided the information that was used as part of the testing. As we discussed with the committee before, the purpose was to understand how people engaged with the guidance. Two options were used in the rehearsal. We considered the results that came back from that rehearsal and we made our recommendation, which is in the report that we provided to the committee.

The letter that we received from the cabinet secretary in relation to the guidance describes

“an ongoing process until the final version is agreed by the end of summer this year.”

That refers to all the guidance.

Are those two options for question 3 still live for discussion, or has a decision been made that the proposed guidance will be the guidance that will accompany the census?

The guidance generally will not be finalised until the summer.

What about the guidance on the sex question, in particular?

The proposal for the guidance is as set out and as I communicated to the committee at the beginning of January. I thought that the committee was aware of that.

What is the purpose of the alternative proposed guidance? It is being presented as though it is a choice.

I apologise—I do not have before me what you have been sent by the NRS, so I cannot answer that question.

That is why I initially addressed the question to Mr Whitehouse.

Two types of guidance came to the committee, which showed what was part of the research. In our report, and in our overview report, we made clear what our recommendation is, which was the basis of the conversations that we had when colleagues and I were here earlier in January.

The guidance for the two versions of the sex question that were tested was supplied to the committee by the NRS in August 2019.

Our version is dated September.

The recommendation report was published in December.

So, it is done.

That was the point that we were at when we provided the paper question set and guidance to the committee. We undertook further testing and development, and the recommendation report, which set out the recommendation for the guidance for the sex question, was published in December 2019.

So, it is done and there will be no further testing of the two options. You have decided that it will be the first option.

All the guidance will not be finalised until the summer. With regard to the evidence base and research, the committee wanted us to look at what would happen and how people would respond to the guidance, so the two options were tested at the end of last year. The report on that testing has been presented to the committee, and the recommendation was based on all the evidence from both the rehearsal experience and the research that took place in December. Although there are issues around legal sex and concerns around self-ID—you can see that in the research report that you have already had a chance to discuss—the recommendation is that, to ensure that we get the best quality and quantity of information, that is what should be in the guidance.

We conveyed the current situation that I am proposing to the committee in a letter to you back at, I think, the beginning of January. If that has not been evident to the committee, it is certainly evident in the report—we can point you to the sections where it talks about that. I would have hoped that the committee would have had a chance to go through the report that was presented to you. It said what would be the best guidance to maximise what we are trying to achieve, which is completion of the census and the best-quality data. We have been very open and we have given you the information.

I understand why it has been described as proposed guidance, because the final decision will not be made until later this year. However, to the committee, having the word “proposed” in front of it suggested that it was still up for discussion. This morning, you say that it is no longer up for discussion—you have done the testing and the evidence directs you towards that decision. Although the guidance is called “proposed”, the debate about it is over. You will not consult or test different options—that is the decision. It would be helpful to get some clarity on that.

The additional testing that was done at the end of last year was over and above what has been happening in ONS and elsewhere. Nothing in that research led us to think that there was a need for further testing. Had it not been clear—had the results of that research been different—there might have been a case for further testing, but that was not evident from the report.

As a committee, you have already had a chance to discuss the report. There was nothing in the report to suggest that we still have concerns about the content of the research or that we have further testing to do because the evidence is not clear. Anybody who looks objectively at the report will be steered towards the conclusion that the binary question with self-identification would give the best result, and that is what I relayed to the committee.

Going back to the point that the cabinet secretary made to Alexander Stewart, ultimately, it is a question of maximising the quality and quantity of the data that is collected. As the cabinet secretary who is responsible for that, are you confident that the approach you propose would do that?

From the evidence that is before me? Yes.

Great. Thank you.

Before we move on, I have a supplementary question that relates to my earlier question about non-response rates. Page 31 of the ScotCen research points out that, in the general population sample,

“For the self-identification sex guidance, 2% said they would ‘skip the sex question’ in the census if the guidance was used and 1% said they would ‘not complete the census at all’.”

The figure was the same for the legal sex guidance.

People cannot skip the sex question. If they skip the sex question, they cannot complete the census, which means that 3 per cent of people would not complete the census. When I raised the issue with you earlier, you said that we do not know why that was the case; however, on page 30 of the ScotCen research it is clear. Under the heading “Reasons why some people find the self-identification sex guidance unacceptable”, it says that

“a common theme among those that found it unacceptable was that sex is binary and determined at birth and that consequently there should be a legal basis to any response to the sex question. For some whom had objections, the self-identification sex guidance was viewed as ‘encouraging’ the notion that sex is a personal choice.”

Contrary to what was said earlier, your report, on which your decisions are based, made it clear that people who said that they might not complete the census because of the self-identification guidance were clear about why that was.

10:30  

That figure is from the general population set that you tested. Given that you have repeatedly said that response rates are important to you, if 3 per cent of the general population say that they will not complete the census, that is more significant statistically than a small number of the trans population having difficulties in completing the census. Your decision was based on that much smaller group as opposed to the general population. The research was done before the issue had become very prominent in the public eye, and 3 per cent is a significant proportion of the general population who might not complete the census.

The ScotCen research was based on people seeing everything, whereas the rehearsal was based on what would be normal domestic completion, which is online at home, without access to the guidance—unless people chose to access it. The statistics were the same for the self-identification and legal sex guidance.

I will ask the statisticians to help me on this, but, despite people saying that they would skip the question, as you said, they cannot skip it, so we cannot say definitely that 2 per cent would skip it. When they started to fill it in, they would realise that they could not skip it, because it is mandatory, so we cannot add the 2 per cent and the 1 per cent and say that they all have the same reasons. You are right, though, that some people said that in the research.

But their reasons for wanting to skip it are clear.

Some are clear. Some people will not complete it for a number of reasons. That happens all the time with censuses.

Your research does not say that. The research says that they object to the guidance because they believe “that sex is binary” and they think that the guidance encourages the view that sex “is a personal choice”.

To ensure that people complete the census, we are putting it online, the questions are mandatory and people will not be able to skip them. The research does not say what happens when people are presented with the issue of not being able to continue if they skip a question. Does that mean that they will continue with it? Will they fill it in or will they not fill it in? We cannot read across on that basis.

Is there anything to add from the statistical point of view?

Around 3 per cent of the general population said that they would not respond to that question, no matter what the guidance was—whether it was the self-identification guidance or the legal sex guidance. The version of the guidance does not change that group of people who said that they would skip the question.

Yes, but they were shown only two versions of guidance, which related to legal sex and self-identified sex.

Yes.

The cabinet secretary alluded to the fact that there is a lively debate on the issue and that people feel strongly about it on both sides. That is clear. Have you considered the impact of people from among the general population boycotting the census because of the guidance?

There is no indication that that would happen. People are conscious of the civic responsibility of completing the census, and the question has consistently been answered well at high levels. From previous experience, I know that some people do not like completing the census, so they will not complete it, and there has to be sensitivity in the prosecution of those cases. In the past, the Crown Office and Procurator Fiscal Service has shown sensitivity to people who, for whatever reason, have a general objection to filling in the census. However, that is an extremely small number of people. I will bring Pete Whitehouse in on that.

Over the next few months, up until the census date, we will be doing a lot of work to promote the benefit of the census, explaining why it is vital for delivering services, developing funding formulas and enabling local charities and third sector organisations to do their work. We will promote the benefits, and we hope that all people across Scotland will be able to engage in the census and fill in the form.

When we look at some of the other issues that have been discussed, there is a need to triangulate all the information. There is no one piece of absolute evidence; we look at past history, rehearsals and the fact that there is a lot of evidence that the vast majority of people can answer the sex question and do so quite happily without any concern, although there is a need to provide some guidance. It is also clear that people have and express different views in a test situation. We absolutely welcome—

They do have different views, but you have based your decision on the response rates of a very small percentage of the population. You have based your decision on how transgender people feel about the question. You have dismissed the fact that a larger group of people in the general population may not complete the census. You are dismissing them.

You cannot compare those two figures. We have evidence from the 2011 census and from our rehearsals that the question is answered very well. People answer the sex question—there is no doubt about that. From our rehearsal, we also know that some people have questions about how to answer it, and that is why the guidance is there. The figure of 3 per cent, from the rehearsal, is useful. It tells us that we have to promote the importance of the census and encourage people to understand why we are asking the question in such a way, that it is consistent with the 2011 census and that it is consistent with the question in the rest of the UK, because that will give people the wider context. We hope and fully expect that people will be able to engage as they always have done.

Remember that, in rehearsal, 99.5 per cent of people answered the question without looking at the guidance, and they completed it not on a self-identification basis but on a male and female basis. They were doing it on a self-completion basis, but they were not accessing the guidance.

In light of the previous discussion, things are a little unclear. The committee has before it the text of the guidance on self-ID for the mandatory sex question, to which you have now signed up. Is that the final text? Alternatively, is it part of the overall guidance that will be subject to further change?

The answer is the latter.

What changes will be made to the guidance?

It is our proposed—

I will rephrase my question so that you can better understand why I am asking it. You say that you have made the decision and produced the guidance on the basis of the evidence; the convener suggests that maybe some other evidence has been given less weight. You have made your decision, so presumably there will be no further testing. Is that correct?

The testing that we did was exceptional and would not necessarily have happened, but it has been helpful.

Okay—there was testing but not a note of guidance or heads of test. In any event, there is no further testing, so what would cause you to change what you have in front of you? I am just not following the process.

Well—

We have done that engagement.

I am just asking whether it is the final text now.

From our perspective, yes. For that sex question—

What does “From our perspective” mean? Who would change it?

The guidance is separate, as I said right at the beginning. The committee’s legal responsibilities relate to the order, which sets out the subjects to be included, and then the regulations, which will set out the questions. That is the main focus. There has been a strength and an openness in discussion because, as the convener has said, it is such a live issue. I know that the committee has become heavily involved in the guidance issue, but its legal responsibility relates to the order.

That is the case but, as I said, Mr Matheson highlighted quite rightly that the guidance sets the legislative context for the question. However, I will not go over old ground again.

That statement relates to a different session.

Thank you very much.

We have received further correspondence from the Sikh Federation UK regarding the issues that it raised and its desire to have Sikh as an ethnic option on the census. The organisation is also having a dialogue with the ONS and has raised the possibility of a legal challenge in the Court of Session in Edinburgh if the ethnic option is not available on the census. Will you provide some information about any dialogue that you have had with the Sikh Federation UK regarding the census in Scotland?

As the committee is aware, my understanding is that the National Records of Scotland met Sikhs in Scotland on 25 October. It has also engaged with a number of individuals and organisations, including the new Gurdwara Glasgow, Sikh Sanjog, Sikh Federation UK, Preet Gill, who is chair of the UK all-party parliamentary group on UK Sikhs, the Edinburgh Gurdwara, the Ekta Group, British Sikh Council UK, the Aberdeen Gurdwara and Lord Singh of Wimbledon.

This is obviously a very live issue—some Sikhs want to be able to answer the question not only on a religious basis but on an ethnic basis. As with other issues, there are different views within the Sikh community, and we know that some Sikh respondents found inclusion of a religion in the ethnic groups unacceptable, inappropriate or confusing.

I have asked the NRS how somebody who wants to answer that they are a Sikh as an ethnic option would do so. If that is what they want, they will be able to write that in. If the Sikh option was to appear as a tick box in the ethnic groups, that might cause difficulties with other members of the Sikh community who find that inappropriate, unacceptable and confusing. Sikh will appear in the religion section as a tick box, but it will be possible in the ethnic section to write in “Sikh”.

With regard to on-going discussions, the Sikh Federation UK has also written to me and I have agreed to meet it.

That is very helpful. Clearly, the Sikh Federation UK is very strong in its belief and it has referred to a case that took place in the House of Lords in 1983, Mandla v Dowell-Lee, as part of its argument. If further legal action were to take place in England, or if legal action were to take place here in Scotland, would that have the effect of delaying the census in Scotland?

It is very serious. From a legal point of view, we are keeping a close eye on what is happening in relation to the other case. With regard to the order, I have set out what we require to do to meet the timescales to build the digital infrastructure, so it would cause a difficulty.

Would there be a financial implication as well?

Yes.

I have a further question on a separate issue. It is more of a technical question on an issue that I have raised in the past on behalf of organisations that represent and work with people who are disabled, and it arises in particular because this census is primarily an online census. I know that a dialogue took place, but is it continuing with organisations such as Dyslexia Scotland and RNIB Scotland?

I will ask my colleagues to answer that question.

Yes, the conversations are on-going. Part of the census is a programme of working with all groups in society to understand what public assistance would be most appropriate to support everybody to complete the census. Those conversations are on-going.

That is helpful, thank you.

I refer to the letter from the Sikh Federation UK. Apparently, the testing in Scotland was brought into the court case in London by the Cabinet Office. It mentioned that the cognitive testing of seven Sikhs in Scotland had led to the position that the ONS took. The Sikh Federation UK is concerned about and is objecting to that. Was your decision based on the testing of just seven people in Scotland?

No. There was a series of focus groups for cognitive testing, and there were 30 to 35 people in each of those groups. A piece of qualitative testing was done on the religion and ethnic group questions. I do not have the numbers to show how many people took part in that.

You do not have the numbers.

I do not have them to hand, but the research material has been published on the NRS website and it is publicly available.

You say that there are different views within the Sikh community. We have had that letter from the Sikh Federation UK, and earlier we received a letter from a prominent Sikh organisation in Scotland. Nobody has written to the committee to express an alternate point of view. It is said that the decision was made because views are divided in the Sikh community, but which organisations in the Sikh community agree with what is being done?

10:45  

We have proactively engaged with people in our approach to the census. I have listed all the different organisations that there has been engagement with. I cannot tell you which ones said that what is being done is appropriate. However, Sikh being listed as a religion and as something that people can write in the ethnicity section creates a balance. In previous years, it has primarily been considered as a religion, and the efforts to change that mean that it could now also be considered an ethnicity. The options will come up in the predictive text boxes in both sections, so people are not being prevented from choosing Sikh in either one.

Yes. You said that.

It maximises choice, but it does not cause difficulty for those who do not want to list it as their ethnicity. I am not in a position to tell you which organisations said yes or no.

The Sikh Federation UK has suggested that it may take legal action. I guess that, by meeting with the federation, you are hoping to resolve that?

I got correspondence around the same time as the committee did, so I cannot comment.

I am looking for clarification, because this morning Pete Whitehouse—and the cabinet secretary, I think—suggested that the committee was already informed and aware that a decision had been made on the guidance for the sex question. However, the letter that Pete Whitehouse sent us on 18 December, which was prior to his appearance before the committee on 9 January, says:

“The current sex question testing, which NRS commissioned ScotCen Social Research to carry out, is progressing well. I will share the results of that work with the Committee as soon as possible. I understand a session has been set with the Committee for 9 January 2020 to consider this work”.

The letter did not tell us that a decision had been made on what guidance to include for the sex question. I would have to look back at the Official Report, but I cannot remember that being said.

I am quite happy to look at what the correspondence said or did not say. However, in January, I was under the impression that the committee had been informed of the decision, which was taken as a result of the research that had taken place at the end of last year.

That is helpful, but the letter did not state that a decision had been made and my memory of the session on 9 January is not that the committee engaged with it as an issue for discussion. I do not think that it has been made clear to us that a decision had been taken.

That is obviously a misunderstanding.

I accept that. Thank you.

Alexander, would you like to ask your question?

My question moves on to a different topic, but perhaps you would prefer to stay on the same topic.

Does anyone want to come in on this question?

I would like to stick to the topic of religion. I wonder how important question 21—the religion question—is. I note that, in the 2001 census, 14,052 people described themselves as being Jedi. That is more than the number of people in Scotland who said that they were Sikh, Jewish, Pagan or Hindu. It is a voluntary question, of course, but 5.5 per cent of people did not put a religion down in 2001, and 7 per cent did that in 2011. I also notice that 291,000 people chose “Other Christian” in 2011—that is four times the number of Muslims, for example.

If there is going to be a religious question, this one seems very vague. As an example, if a person is agnostic or atheist, they are stuck between identifying with one of the religions that are listed or choosing none. If you want accuracy, surely, the question should be whether someone considers themselves to be atheist or agnostic and then a list of denominations.

Lastly, why is Episcopalian not listed? There are more Episcopalians—370 congregations—than Sikhs, Buddhists, Hindus, Jews or Pagans. In fact, there are more Episcopalians in Scotland than Sikhs, Jews and Pagans combined.

I would like to hear the thinking behind the question, and I would like to know why it is so vague.

It is not the final text of the question. Some points that were made previously—by, I think, Ross Greer—have been taken on board for change.

It is a big and fairly fundamental question. That is exactly why the previous committee, at the previous census, thought that it would be a good idea to have early engagement with the committee, before the order was placed. Obviously, the order that is in front of you refers simply to “religion”; it is a catch-all, and you will have to decide whether you agree with that. However, I hope that the process that we have gone through will have been able to flush out some out of those fairly fundamental, bigger issues, because it is quite late in the day to necessarily address that.

Kenneth Gibson’s point about Episcopalians is reasonable. Years ago, as part of the topic guide, I am not sure whether Episcopalians came up as an issue that people thought there was a need for a question on; perhaps officials could give you more information on that. However, it is kind of a big question to be considering at such a late stage, if you see what I mean.

As the cabinet secretary said, there has been some slight adjustment to the religion question following conversation with Muslim groups, who asked for certain variation on that question to allow them to write in more fully. There is a writing box that allows that to happen, and there will be predictive text that allows people to do that, which helps with our processing.

The question itself is a fairly routine one that is asked in other surveys and which provides what data users want. If a data user is particularly interested in a particular aspect of one particular religion, the predictive text or writing boxes will provide that analysis. The question works, and it is used regularly. We have made some slight adjustment to encourage and enable full participation, and there is a writing box that allows that data to be processed quickly, and data users to do the more detailed analysis that they wish to.

Episcopalian people would therefore write “Episcopalian” under “Other”, and that information would be gathered and published as part of the reports. It is simply not there as a tick-box question.

I am not an Episcopalian by heritage myself. However, given the number of Episcopalians that there are relative to the number of some of the other groups who are in there, I would have thought—on the basis of reasonableness—that they should be in there.

However, the main question for me is, if I was an agnostic, what box would I tick?

I think that you would tick the box “None”.

But “None” does not mean that you have no religion. You could be an agnostic but have a Presbyterian or Roman Catholic background. All I am saying is that the question could be worded a bit better such that it asks people what they consider themselves to be—agnostic and atheist are completely different, for example. Agnostic obviously means that you are not sure, whereas atheist means that you do not believe. In addition, a lot of agnostics might identify with the religions that they were brought up in. Why ask a question if you are not looking to be more accurate on it? This question is pretty vague.

I suppose that it is about census use. Obviously, particularly for minority religions, part of that is about society’s responsibility to ensure that there are services and support and so on. However, for agnostics and atheists, I am not sure what the service use would be. There might be a kind of capturing of society and of what people believe, which you can see the demand for from an academic point of view. However, has there been demand from agnostics?

What is the service use for pagans, then?

In the development of the 2011 census, the Scottish Pagan Federation approached the GROS, as it was then, to consider the inclusion of a pagan question in the census. A large amount of work was done at that point, although, following the parliamentary process, it did not get included. We were approached again for 2021. There is an identified need for information on the pagan community in Scotland, and we are responding to that. Any changes that we propose for 2021, for any of the questions, are all based on the fact that data users are coming to us and presenting a clearly identified data need. We then prioritise those needs, because there is obviously a limited amount of questions that we can ask.

Okay. Thank you for letting me know that there is a pagan community in Scotland, because, despite living here my whole life, I was not aware of that.

That is why the information was shared with the committee over many months. Although I know that there has understandably been a lot of focus on the sex question, all those other issues have been lying there to be engaged with. However, we are at the final part of the process.

I want to ask about the health questions—that is, questions 17 and 18. Question 17 talks about the general health of individuals and gives a bit of guidance as to what that is. Question 18 talks about specific types of illness that individuals may have, and there are about three pages of guidance on that.

You talked about the guidance not having been used very well by a number of individuals when it came to the sex question. When you were testing questions 17 and 18, how many people used the guidance to try to find out what their health situation is? You have listed dozens and dozens of ailments that individuals may have, but people might not see them as being ailments that they have unless they click on the guidance.

One of the bits of information that came back to us following 2011 was that, although the questions gathered the information that meets the user need, there was a large volume of write-ins, all of which had to be coded back, which was a time-consuming process that did not come without a cost. Given that we knew that it would be digital first, the development work for the 2021 census was about how to present the material so that users who have a condition would be able to answer the question easily.

The question is formatted slightly differently online. On page 52 in the version that you have, there is the heading

“Deafness or partial hearing loss examples”.

The online version has a little link under that, so the guidance does not come as one.

So the guidance comes as each specific question materialises. If someone feels that they have a mental health issue or a long-term issue, would there be specific guidance to cover that?

Yes, that is correct. They could access just that list of conditions—

Rather than them all.

Yes. When testing the online functionality and guidance in the development of the question, we found that respondents preferred that, because they could easily see where they should go. In addition, a main point is that a person might not think of their condition as a mental health one, so the guidance helps people to understand.

Exactly. I see that you have listed blood pressure as a specific long-term issue. However, although you are identifying it as such, individuals who have that condition may not perceive that they have a long-term condition. It is the same with regard to mental health; if a person has panic attacks or anxiety, they may not see that as a mental health condition. To go back to my original question, when you were testing, did the majority of people look at the guidance?

No. The majority of people have no condition and do not need guidance. The guidance is there only for those who have a condition and who need guidance as to where that condition should be grouped.

However, as I have identified, they may not perceive that they have a condition. You have listed blood pressure as a long-term condition, but someone who has a blood pressure problem may not identify it as a long-term condition as that sounds much more serious and dramatic and like something such as cancer. If people do not click on the guidance, you may not get the right information.

People will use the guidance if they feel that they need guidance; that is the same for all the questions. Although I take your point, in the testing using the online functionality for the question, some people used the guidance. Obviously, with rehearsal evaluation, the number of people who access the guidance is quite small.

It is easier for people to identify the guidance in the online question, but what about those who are using a paper copy? If they are filling in the question as it stands, how will they identify the guidance for each category?

There is guidance that is tailored specifically to the paper version of the questionnaire, which is also available online, in a different place from the online questionnaire.

As Stuart McMillan said, people may not use the online questionnaire because they have a learning difficulty or disability, and they may have someone supporting them to ensure that you get the correct information. The guidance will be vital to such individuals to ensure that the information comes through. You could miss out on some of that if it is not identified and captured correctly.

I understand what you are saying. In the testing, we found that having the online guidance helped people to categorise themselves into one of the high-level categories, which reduced the number of write-ins. Other than that, the question is largely the same as it was in 2011, but I appreciate that not everybody will read guidance and answer accordingly. The list is not exhaustive, but those are the categories that came up the most. We have worked with other existing lists, such as the World Health Organization’s list, on how to categorise, but I take your point.

11:00  

The situation will be similar to that in 2011, when it was all paper based and people who needed additional support for anything would either phone up or get help from one of the enumerators. They could also make contact by email. If anything, because the 2021 census is digital and online, more people, although not everybody, will be able to get the support that they need. I am conscious of the point about special support for people with disabilities, but that was also the case in 2011 and 2001.

There are a couple of new areas that I want to quickly wrap up. There may be an easy answer to this question. Article 7 of the order is about

“Disapplication of liability to penalty for refusing or neglecting to state particulars with respect to”

the transgender question, which of course is voluntary. My understanding was that the sexual orientation and religion questions are also voluntary, but there is not a similar provision about those areas. Am I missing something?

The Census Act 1920, as amended, makes those voluntary. No further subordinate legislation is required to do that.

Okay. Was that when we amended the act previously? I thought that we amended it to make the transgender question voluntary as well.

The Census (Amendment) (Scotland) Act 2019 leaves it in the position that there needs to be an exercise of subordinate legislation powers to make the trans question voluntary.

Is that distinct from the sexual orientation question?

The sexual orientation and religion questions are voluntary, because that is in effect hard-coded into the primary legislation. It was left to an exercise of subordinate legislation powers to make the trans question voluntary, so the 2019 act paved the way for getting to where we are.

Fair enough.

Remember that we separated the two, and that was the consequence.

So we did. That is fine. I thought that it would be better to check that.

Finally, another area of debate that the committee had was when we discovered that the question on sexual orientation, which we agreed should be voluntary, includes options for other sexual orientations. We had a long discussion with the NRS around the fact that sexual orientation is clearly defined in the Equality Act 2010 and we wrote to the NRS about that. The census order continues to ask for other sexual orientations and I understand that there are predictive response options, which the ONS is not using and which are very unfamiliar to some people.

There are two objections. One is that it suggests that there is another sexual orientation apart from gay, straight and bisexual, which is basically what the 2010 act says, and that causes confusion. The second objection is to the predictive text. In that regard, Dr Kath Murray of the University of Edinburgh said:

“A sexual orientation question with overly complex predictive response options, not all of which describe the sex that a person is attracted to means that less complete information may be collected on the protected characteristic of sexual orientation”,

which is the reason for asking about sexual orientation in the first place—it is a protected characteristic. Dr Murray went on to say:

“For instance, ‘demiromantic’ does not tell us whether a person is gay, lesbian, bisexual or heterosexual,”

so that answer would not give meaningful information. Do you still plan to go down the route of those predictive texts?

I have some sympathy with that point. There should be an option for “other”, that people could fill in, as there is for lots of other questions right across the census. I am less convinced that we should pursue the predictive text, so I am considering that.

I am sure that the committee will consider that answer. Thank you very much for giving evidence.

11:05 Meeting continued in private until 11:28.