Losing consensus on the census

Within the last fifteen months, an intense debate has broken out over whether the sex question in the 2021 census should be framed in terms of self-declared gender identity, legal or biological sex.

We think that this loss of consensus relates principally to the approach taken to question development by the census authorities, whereby the sex question and guidance has been effectively co-produced with a small number of transgender advocacy groups, without wider consultation.

As a result of these dynamics, the current National Records of Scotland (NRS) recommendation on the sex question, which supports framing the question in terms of self-declared gender identity, is weighted towards a limited set of respondent interests. We would suggest that this format is not in the wider interests of data users, including policy-makers and researchers, and is likely to weaken our ability to address sex discrimination.

The loss of consensus on the sex question has also increased the likelihood of a higher non-response rate, compared to previous censuses. Testing by ScotCen shows that this is likely whether people are guided to complete the sex question as self-declared gender identity or as legal sex.

We would suggest that the best option to minimise this damage is to revert to a question without guidance, as was the case in all censuses between 1801 and 2001.

Question development
1.  Our work suggests that the relationship between the census authorities and transgender advocacy groups have developed over at least a decade at UK level (the 2011 gender self-identification guidance for the sex question was initially developed by ONS). We have found no evidence of wider consultation or transparency in relation to the introduction of the 2011 guidance. For example the 2008 NRS report on its recommendations for the 2011 census states that no changes will be made to the sex question, and makes no mention of the introduction of gender self-identification guidance. The 2011 EQIA also makes no mention of the new guidance.

2.  During the question development phase for the sex question in the 2021 census, NRS met only with LGBT advocacy bodies. There is no evidence of consultation with independent statisticians or census data users in this period (see FOI correspondence). We would suggest that awareness of the proposed guidance and its implications remains poor within the relevant academic community (for example, administrative data users).

3.  Minutes from the Scottish Government’s Board of Official Statistics indicate that the Board had no input or discussion on the 2021 Census prior to the Census (Amendment) (Scotland) Bill being laid in Parliament. Ten days after the introduction of the Bill, on 12 October 2018, NRS updated the Board on its plans, although it is not clear in what detail, or if the proposed changes to the sex question were discussed

4.  Further FOI correspondence between the NRS and the Equality Network suggests an unusually close working relationship between the two organisations.

5.  In earlier correspondence between ourselves and the former Head of Statistics at NRS (at the stage where a non-binary sex question was still under consideration), it was stated that there was a respondent need for a non-binary response to the sex question; but that no one had requested the data from an outputs perspective. The correspondence also stated that random imputation methods would be used to assign non-binary data to binary categories at the stakeholders’ behest, who requested that imputation should not be based on any other characteristics (as is standard practice).

6.  The terminology used by NRS reflects a distinctive agenda associated with transgender advocacy. This includes terms such as ‘assigned at birth’[i] and ‘cisgender’.[ii] These are not neutral descriptive terms, but are underpinned by particular ideological assumptions that many people reject (namely that everyone has an innate gender identity, and that for some people this does not match their biological sex).

Transparency and engagement
7.  We would suggest that the level of transparency and engagement regarding the development of the sex question remains weak.

8.  In September 2019 a group of Scottish academic population data users wrote to the CTEEA Committee to express their concerns in relation to data reliability (see here). NRS met directly with some members of the group to discuss their concerns further, however NRS have not taken these on board in its final recommendation. NRS minuted this meeting, although the account was challenged by Professor McVie who noted that several key points had not been documented. A copy of the NRS minutes and Professor McVie’s comments and arguments can be accessed here:

9.  NRS published its report on the sex question on Friday 20 December immediately before the Christmas break. Despite significant public interest, this was not distributed via its public newsletter mailing list. A Census Newsletter was issued on 19 December, however this made no explicit reference to the imminent release.

10.  NRS state that question development and response options are evaluated against the strength of user need. However, its recommendation on the sex question discounts the views of 80 senior academics working with population data, including professors of statistics and ten Fellows of the British Academy who have called on the census authorities not to conflate sex and gender identity.

11.  The note from 80 academics was co-ordinated by Professor Alice Sullivan who in a recent social media post stated: “I organised a letter to the census authorities about plans for the sex question. I received several apologetic emails from colleagues who were too frightened to sign this letter”. This culture is evident within universities. We also understand that debate on the definition of sex and gender is being stifled within at least one major research organisation, and think that there is a high risk that the same culture exists within government and other public bodies.

12.  The NRS sex question recommendation report states that it believes ‘a binary sex question asked on a self-identification basis provides the best balance in meeting the diverse range of user needs across the full census dataset’ (p.32). However NRS have refused to provide information on data users requiring data based on self-declared gender identity, apart from to suggest this pertains to individuals and not organisations.

CTEEA Committee session 9 January 2020
13.  NRS gave evidence on the census order to the CTEEA Committee on 9 January 2020, where the discussion focused on the sex question guidance. The Official Report can be accessed here.

14.  The issue of academic opinion was repeatedly raised, in particular, on whose views NRS had based its recommendation for a self-declared gender identity question, and the respective relevance and expertise of different views.

15.  Mr Whitehouse, NRS Director of Statistical Services, refused to disclose the details of any census users who had requested data based on a self-declared gender identity question despite being pressed several times by Committee Members.

16.  Committee members also pressed Mr Whitehouse on why NRS had not taken into consideration the relevant expertise of the 80-signatory letter advising against guidance for the sex question based on self-declared gender identity. These exchanges revealed that NRS had instead placed greater weight on a letter organised by a pro-self-identification researcher, signed by a group of 53 people working in higher education, few of whom provided details of their area of academic expertise. It is not clear what due diligence NRS has done on this letter, but we can identify only a handful who appear likely to be quantitative researchers in a relevant field: most are from backgrounds which do not imply relevant academic expertise on this topic (including quantum physics, theatre studies and chemistry).

The Convener: I am sorry to interrupt, but I just want to drill down on that. Which data users have asked for a self-identification question because of their wish for that data as an output? You have dismissed all those senior academics, such as Professor David Bann, associate professor in population health at University College London, and Professor Mel Bartley, professor emerita of medical sociology—I could list many of them— who use population data. Which data users wish for the sex question to be self-identified?

Pete Whitehouse: I am keen not to get into individuals and saying that people have a particular—

The Convener: But in your letter to me, you say that you have had meetings with data users, so who are the data users who want that change?

Pete Whitehouse: We had representations in letters from a number of users.

The Convener: There were other communications, but they were from academics in other fields. Many of them were from fields such as computer science or literature. I am sure that they are sincere in their views and expert in their areas but, unlike the academics I am talking about, they are not all users of population data, such as sociologists, social scientists and economists. They were a different group of academics.

17.  Later in the meeting, the Convenor stated:

The Convener: You keep saying that data users have asked you for clarity and that they want guidance. However, to go back to my original question, I talked about Alice Sullivan’s letter and the 80 academics, including Professor Susan McVie, with whom you have engaged, who are all concerned about prominent digital-first guidance that erases biological sex as a characteristic. You have dismissed them, but you are not telling me which data users have requested that clarity and that self-identified gender identity be conflated with biological sex.

Pete Whitehouse: The committee’s papers include representations from groups of academics. We listened to those academics, and they have different views.

The Convener: No. The academics who use population data are very clear that they want a biological sex question. Are you saying that you have dismissed them to listen to another group of academics, most of whom are not social scientists who use population data and who might include, for example, professors of literature, of queer legal studies or that sort of thing? I am sure that they are experts in their fields, but they are not social scientists who use population data.

18.  MSP Kenneth Gibson also stated:

… we do not know who the people are on the other side of the argument. Those 80 academics suggested that the approach is not appropriate, but we are not getting hard information as to who the people who have a different perspective are. We are hearing views, but not who is putting forward those views.

19.  At the end of the session, the issue was raised again, in an exchange that highlights the crucial difference between data users and census respondents.

The Convener: Okay. To finish, I want to go back to the issue that a number of members have raised with regard to who the data users were that had asked you for self-ID guidance. You said that you did not want to name names. Are you talking about the submission to the committee of 20 September 2019 from a number of academics—I think it was around 50—who wrote in support of self-ID guidance?

Pete Whitehouse: If that was the one with, as you say, around 50 signatories, then yes. That is one of the groups of people who have expressed—

The Convener: We know that you have been lobbied by lots of stakeholders and campaigners on behalf of the community of relevant respondents, but that is a different thing from independent data users. If we talk about data users that you have responded to, are we talking about the academics who signed the letter of 20 September? Are those the people you are referring to?

Pete Whitehouse: If you are asking for full details of all the academics who have been in touch with us, we can come back to you on that.

The Convener: It is not just academics, though, it is data users who are independent researchers and are saying that they want a biological sex question, a legal sex question or a self-ID sex question. You have said that you have been swayed by data users who favour self-identification. However, you were not able to tell us who they were. Are you talking about the 50 or so academics who signed the 20 September letter?

Pete Whitehouse: Those are some of the people who expressed a need for that.

The Convener: In contrast to the letter from Professor Alice Sullivan, which is signed by senior academics—mainly professors—who are social and economic researchers, the 20 September letter is signed by people who are not as senior, and, in most cases, it does not say what their expertise is. I looked some of them up. I will not name them, because that would not be fair, but they include researchers into medieval literature, materials chemistry and computer studies. Is it fair to give weight to that group of people? They are not data users; they are just academics who feel strongly about the issue, as opposed to Professor Sullivan’s group of 80 senior social and economic researchers, experts in medical sociology and people such as Professor McVie, who sits on the Government’s advisory group on statistics. I am confused as to why you are favouring that other group of people, who do not have that expertise.

Pete Whitehouse: I do not agree with the idea that we are judging and therefore finding certain voices. We are saying that that is how the question has been asked and that is how guidance has been used. We are clarifying that, and there is a need for guidance.

20.  If NRS cannot answer questions on which data users (working in relevant disciplines) are asking for self-identification data, even in general terms (for example, the type of organisation or individual), this is a serious impediment to accountability and transparency in relation to the 2021 census.

21.  MSP Annabelle Ewing asked whether NRS was running ahead of the law by introducing a question that was aligned with the Scottish Government’s proposals on reforming the Gender Recognition Act to allow for the self-identification of gender:

Annabelle Ewing: At the committee’s evidence session on 12 September 2019, the convener herself put to Amy Wilson the proposition that, in referencing self-ID, NRS was seeking to jump the gun—in effect, to usurp the role of the Parliament, which in due course will have an important debate on the matter. NRS’s response to that proposition was not clear, so perhaps I could put it to you again today. Is it the case that NRS is seeking to jump the gun on the self-ID debate, thereby usurping the role of the Parliament?

Pete Whitehouse: From my perspective, we are presenting a question that is well understood, with guidance that helps a group of the population to answer that question. We are not jumping any gun or trying to change or get in front of all the other conversations, discussions and work that the Parliament and others are doing in this space. We are saying that, from 2011, the question has been answered in a particular way and we are providing guidance to support that. In essence, that is where that census work starts and finishes.

22. While Mr Whitehouse stated that this was not the case, the NRS self-assessment report ‘How the National Records of Scotland is ensuring Census 2021 is trustworthy, high quality and of value to users’ notes that census development was aligned with proposals for GRA reform ‘to ensure a cohesive approach’:

‘It is important to recognise that the development of Scotland’s Census sits within a wider Scottish Government context. NRS has worked closely with colleagues in the Scottish Government undertaking the Gender Recognition Act consultation and consultation with colleagues in Equality workstreams to ensure a cohesive approach to the census questions.’ (NRS, 2019: para. 3.29)

23. Ms Ewing also highlighted an inconsistency in approach to different questions, whereby compatibility with the Equality Act 2010 was cited by NRS in relation to some parts of the census, but not to the sex question, despite sex being a protected characteristic:

Annabelle Ewing: …A number of members raised the issue of the interplay between the Census Act 1920 and the Equality Act 2010, which we discussed at some length at the evidence session on 12 September. For example, everybody on the committee was 100 per cent signed up to, and supportive of, the inclusion of the two new voluntary questions on sexual orientation and on transgender status. Those questions were included, further to the policy memorandum, on the basis that their inclusion in the 2021 census was needed to discharge the public sector equality duty. There would surely have to be consistency between what we are doing further to census legislation and the 2010 act across the board. Otherwise, it begs the question of the stated basis for the inclusion of the voluntary questions—which everybody supports.

…Obviously, if part of the census is expressly stated to be informed by the Equality Act 2010, that begs the question about the other parts of the census, where relevant issues exist with regard to protected characteristics. It would be difficult to argue that those other areas are not also expressly related to the Equality Act 2010.

24.  In comments relevant to the loss of consensus on the sex question, MSP Kenneth Gibson stated:

Kenneth Gibson: …it seems to me that, from the start, NRS has had its own agenda on the issue, regardless of what other people think. For example, the convener talked about the letter from 80 academics. NRS did not originally want a binary question. It was only after the evidence was presented and this committee was overwhelmingly in favour of a binary question that it has been changed. Since then, you have evolved to say that a binary question is being used in the rest of the UK, so that is probably a good thing. With regard to the arguments that we have been having about guidance and self-identification, from my perspective—and possibly that of colleagues—it almost seems that NRS is fighting a rearguard action. As an organisation, you have been dragged kicking and screaming into having to ask a binary question on sex.

25.  Mr Gibson also commented on the use of ‘cisgender’ in the NRS report on the sex question, noting this was a contested and politicised term that was not widely understood.

NRS report on the sex question
26.  Professional concerns around self-identification and data reliability are now well-established. The additional comments below relate to the recently published NRS Sex Question Recommendation Report, and focus on the arguments put forward by NRS to support a self-declared approach to the sex question.

27.  The ScotCen trans and non-binary sample is small (75 respondents), self-selected, with nearly half of those were recruited by personal contacts (47%). The ScotCen report states that ‘the findings relate only to those who took part and inferences to the wider trans or non-binary population in Scotland cannot be made’ (p.53). Nonetheless the views of this very small group has fed directly into the NRS recommendation on the sex question.

28.  NRS state that ‘A binary sex question with self-identification guidance therefore supports participation for all people with the census and clarifies to data providers and data users the basis of the question’ (p.7). However, ScotCen testing based on its larger general population sample (2,208 respondents) shows that up to 3% of respondents may refuse to answer a self-identified question on principle (p.3), compared to a 0.8% non-response rate for the sex question in the 2011 census (see page 143). The same proportion stated that they may not answer a legal sex question (p.3).

29.  For the trans/non-binary sample, 21% felt that self-identification guidance was not acceptable, and 77% felt that a legal sex question was unacceptable (p.5). These results show there is no consensus on a self-identified sex question, even within the small, self-selecting sample of the trans/non-binary population, or on a legal sex question. In other words, both approaches now risk increasing the non-response rate. It is therefore unclear why NRS have discounted the approach taken between 1801 and 2001, which is to have no guidance.

30.  NRS state that one of key strands underpinning its recommendation is that ‘It is important to data users to have clarity through the availability of guidance on the basis for completion of the sex question’ (p.7). It is not clear how the proposed guidance will provide clarity to data users, apart from to confirm that the data consists of two different concepts and that it is not possible to disentangle the proportions of each one. Indeed, the guidance risks increasing the number of cases which do not provide data on sex as an objectively defined phenomenon, by directing people to answer on the basis of gender identity who otherwise might not have interpreted the question in that way. NRS have also provided no information on how it will refer to the data (both ‘sex’ and ‘gender identity’ are inaccurate, given the question muddies the two), or how users will be advised to use the data in analysis.


Notes

[i] The phrase “sex assigned at birth” reflects a belief that a prediction is being made at birth about the inner gender identity a person will later manifest, rather than an observation being made of whether they are female or male in reproductive terms. The phrase originates with the treatment of the very small number of people whose physical sex is ambiguous at birth, estimated to be around 7 or 8 cases in the UK per year. See: https://www.parliament.scot/S5_European/Inquiries/CensusBill_DSDFamilies_CTEEAS518CB33.pdf.

[ii] ‘Cisgender’ is used to refer to a person who is believed to have a sense of gender identity which aligns with their biological sex. The term is contested and rejected by those who critique the underpinning assumption of innate gender identity.