In February 2020 the Scottish Government confirmed Scotland’s next census will advise respondents to answer the question in line with their ‘lived sex’.
However, in England and Wales, which is conducting a census in 2021, debate continues about the form of guidance that will accompany the sex question. This has presented opportunities for several myths to spread about the design and purpose of censuses in the UK.
This updated guide (previously published in January 2020) addresses these confusions head-on and provides seven helpful explainers that cover the use of census data, the alignment of the census with the 2010 Equality Act, international practice, guidance provided in previous censuses and what is meant by self-identification.
New Zealand’s official data agency has proposed a ‘gender by default’ approach to how the government collects data about sex and gender.
This proposal is part of a consultation being conducted by Stats NZ, New Zealand’s national statistical organisation and the body responsible for the country’s 2023 census. The consultation describes ‘gender by default’ as an ‘approach that defaults to the collection of gender data as opposed to sex at birth’ and that ‘collection of sex at birth information should be viewed as an exception where there is a specific need’.
As I have noted in previous blog posts about the design of the sex question in Scotland’s 2021 census, Stats NZ argue that ‘in most cases a person’s gender – their social and personal identity – is most relevant for policy making and research rather than their sex at birth. Gender based analysis is used in a range of areas, from income equality to health and education.’ This also echoes guidance from Engender, a feminist policy and advocacy organisation based in Scotland, which advocates for better quality gender-sensitive sex-disaggregated data.
The Stats NZ consultation is in response to inconsistent practices in the collection of data about sex and gender, ambiguity as to how respondents should answer questions about sex and gender, and the need for more inclusive practices, particularly for trans respondents and Māori and Pacific people (which have culturally specific identities related to sex and/or gender).
Applying a gender by default approach to data collection
To apply a ‘gender by default’ approach to data collection, Stats NZ outline a three-step guide for those looking to collect data about sex and/or gender (see the image below).
Firstly, and routinely overlooked, data collectors should ask whether they need to collect data about an individual’s sex and/or gender. Rather than assuming that ‘more data’ is always a good thing, data collectors should assess the potential benefits of collecting this data and ensure this outweighs any potential negatives.
Secondly, data collectors should ask if is it necessary to distinguish between transgender and cisgender respondents. If not, as is likely the case in most data collection exercises, the following question about gender is recommended:
In data collection exercises where information about trans and cis populations is required (such as certain health care contexts), Stats NZ recommends a two-step question where a question about sex at birth is followed by a question about gender.
Thirdly, data collectors should ask if they need data about the intersex population (an umbrella term that describes people with a wide range of variations in sex characteristics). If so, this will not be collected by the previous two questions, and will therefore require an additional question that specifically asks if the respondent was born with a variation of sex characteristics.
The proposed question on gender, which would be asked in the vast majority of data collection exercises, is unambiguous in what it is asking and is a departure from use of the term ‘gender identity’. Furthermore, it addresses the perennial problem of the ‘other’ response option, which is an inappropriate way to ask people to describe how they identify. The use of the term ‘another gender’, with the provision of a write-in box, presents a model that other data collectors may wish to replicate.
Gender by default – the future for data collection?
No approach is ideal and, as people respond to the consultation, it is likely that Stats NZ will make further tweaks and changes. However, for those developing guidance on the collection of sex and gender data in other parts of the world, Stats NZ’s proposal articulates a clear vision that demystifies the substance of the question and is more inclusive for all respondents.
In Scotland, for example, the government’s Sex and Gender in Data Working Group is soon to publish its guidance on the collection of sex and gender data. Stats NZ also notes ongoing research into guidance on the collection of sex and gender data in Australia and Canada.
Those working on the design of sex and gender questions are at risk of losing sight of the research questions we actually wish to answer when we collect data about sex and gender. By bringing us back to the original research question and working outwards from there, New Zealand’s ‘gender by default’ proposal cuts-through the background noise and potentially provides an approach to data collection to roll-out across other national contexts.
As with all other identity characteristics, how people understand and define their sexual orientation is not always simple. The dimensions that contribute to sexual orientation are multiple and, for many people, do not neatly align. They can change over time, differ according to context and intersect with other identity characteristics such as sex, gender and trans identity/history. Finding a census question that can capture this complexity was always going to be difficult.
Even against the backdrop of a global pandemic, preparations for Scotland’s 2021 census continue. On 21 March 2021, the Scottish government will ask everyone aged 16 and over, roughly 4.5 million people, to voluntarily disclose their sexual orientation. For the first time, Scotland’s census will present the opportunity for respondents to identify themselves as ‘Straight/Heterosexual’, ‘Gay or lesbian’, ‘Bisexual’ or ‘Other sexual orientation, please write in:…’ in a national data collection exercise. On the same day, censuses will also take place in Northern Ireland, England and Wales, asking respondents to disclose their sexual orientation.
Predictive lists in the census
One area of interest that has emerged in recent months concerns the use of predictive text for the ‘Other sexual orientation, please write in…’ response option.
To aid analysis of online census responses, National Records of Scotland (the government body responsible for the delivery of the census) proposed using a technology that would predict and auto-populate a response option for people who started typing a sexual orientation in the ‘Other’ write-in box. In a letter to the Scottish Parliament’s Culture, Tourism, Europe and External Affairs Committee (designated as the lead committee scrutinising legislation related to the census), NRS explained how this approach would:
Improve the respondent experience (it would be easier to complete the census).
Improve data quality (auto-populated options would be matched to a coding list, which would reduce the risk of error from manually matching free-text options to a coding list).
Improve efficiencies in the coding of the data (more of this work could be automated).
Those completing this question would have the option of accepting the auto-populated response or writing-in something different. In the letter, NRS noted that ‘use of predictive text minimises errors such as spelling mistakes and abbreviations, which means clean codeable data is collected’. This meant that the provision of suggested options for respondents who start typing in the write-in box would help maximise the usability of data collected on ‘other’ sexual orientations.
The provision of an ‘Other’ option with a write-in box was welcomed by LGBTQ organisations in Scotland. Equality Network and LGBT Youth Scotland described its inclusion in the census as ‘vital’ and explained that, in their own equality monitoring, between 10% to 20% of the people they engage with identify as a sexual orientation other than lesbian, gay or bisexual.
Following the announcement of NRS’s proposal, media attention focused on the draft list of 21 sexual orientations that would auto-populate the write-in box for the sexual orientation question. The sexual orientations included on the list were drafted with input from LGBTQ organisations and included identities such as asexual, bicurious, pansexual and queer.
Organisations including the Christian Institute, the Catholic Church and LGB Alliance voiced their opposition to this proposal, with LGB Alliance writing to the Committee to note this proposal ‘would suggest that other sexual orientations exist beyond attraction to the opposite sex, same sex or both sexes’. Their opposition went further and urged the Committee not to include the term ‘other sexual orientation’ in the question at all.
Earlier this month, NRS wrote to the Committee to announce that it had taken the decision to remove the predictive text functionality for the ‘Other’ category of the sexual orientation question. The letter also noted that this decision did not apply to other census questions on identity characteristics and that ‘All other Census questions where NRS had proposed to use this functionality will continue’. Correspondence between NRS and the Committee from December 2019 explained that predictive text functionality would be used for census questions on religion, national identity and ethnicity. Although the list of predictive options were still a work-in-progress, this included 116 religions, 274 national identities and 241 ethnic groups. Like the proposed list of sexual orientation options, these lists were based on previous censuses, other surveys, statistical desk based research and engagement with stakeholders.
What does this tell us?
Considering the complexity of religious, national and ethnic identities, what does this decision tell us about understandings of the sexual orientation question in Scotland’s 2021 census? The census can bring into being a population that ‘makes sense’ to the heteronormative majority, yet this risks ‘designing-out’ queer lives and experiences that fail to match these ideals.
Following the lines of scrutiny pursued (and not pursued) by the Committee indicates that members were more interested in a sexual orientation question that ‘made sense’ than the accuracy of data collected. The Committee’s disproportionate focus on questions related to sex and sexual orientation, particularly where they pertained to the lives and experiences of trans people, meant that there was limited discussion about the use of predictive text in questions on religion, nationality and ethnicity. This suggests that concerns expressed about the sexual orientation question were less to do with the census technology deployed and more to do with attitudes towards trans inclusion.
On 1 April 2020, against the backdrop of an international public health emergency, the United States of America conducted its 24th national census — one of the largest data collection exercises in history.
The US census did not ask a question about sexual orientation but it did frame queer identities in a way that endorses a potentially dangerous precedent for future data collection exercises.
The census — which is conducted every ten years — asked only nine questions per respondent, with questions about the identity characteristics of sex, age and race. Data collected by the census is used to determine the allocation of federal funding to states and communities, as well as the distribution of seats per state in Congress.
Although the census did not explicitly ask about sexual orientation, queer identities were not fully absent from the 2020 census. Each respondent was asked how they relate to others in the household, with the question: ‘How is this person related to Person X?’ and presented with options that included ‘Same-sex husband/wife/spouse’ and ‘Same-sex unmarried partner’.
This meant that queer couples, whether married or unmarried, could register their existence in this national data collection exercise. Furthermore, the inclusion of the non-gendered terms ‘spouse’ and ‘partner’ might suggest some acknowledgement of identities and relationships that sit beyond the male/female binary (although this is contradicted by use of the term ‘same-sex’).
In the eyes of the state, who is considered a legitimate subject of being counted?
This falls short of what was initially mooted. Until the election of Donald Trump, the US Census Bureau had planned to add a question to the census on sexual orientation. This proposal was dropped and the Trump administration has subsequently attempted to remove questions about sexual orientation from other federal surveys, including the National Survey of Older Americans Act Participants and research undertaken by the Department for Housing and Urban Development (Loewy, 2017).
Although some might see the inclusion of same-sex partnerships as a small victory, this proxy measure for sexual orientation evokes the ugly question: in the eyes of the state, who is considered a legitimate subject of being counted? The glaze of respectability that might come with ‘same-sex partnerships’ has granted some queer people the privilege of being counted. Yet, a data collection exercise that counts some at the expense of all seems antithetical to a queer political project where ‘respectability’ is a pre-requisite for being counted and queer people play by the arbitrary rules written by a non-queer majority.
Lessons from the US census
The US is not unique in this regard. In their study of large-scale surveys in Europe that captured any data on LGBTIQ people, Karin Schönpflug et al found that only the UK’s national Household Survey explicitly asked about respondents’ sexual identity (Schönpflug et al, 2018). All other surveys reviewed either omitted any reference to sexual orientation or, when mentioned at all, appeared via proxy measures about relationship status (for example, legal couple status or cohabitation as a couple).
These critiques are timely as we are now less than a year away from the UK censuses being conducted on 21 March 2021. Unlike the US census, UK censuses will — for the first time — ask explicitly about respondents’ sexual orientation. In Scotland, the census question is likely to be ‘Which of the following best describes your sexual orientation?’
Although the addition of this question to the census did not face much opposition, particular challenges emerged around the list of response options provided and the ‘Other sexual orientation, please write in’ box. In Scotland, for example, a small number of politicians (none of whom, as far is known, identify as queer) criticised proposals from the National Records of Scotland, the organisation responsible for delivery of the census, to offer more than a narrow list of sexual orientations in Scotland’s 2021 census.
Data from the US census will present a skewed account of the country’s queer population. Any attempt at quantification, using the proxy measure of same-sex partnerships, will under-count the number of queer people in the US and present a false account of queer diversity. Just like other identity characteristics, such as race and disability, measuring sexual orientation is complex as it cuts across inter-related ideas of attraction, behaviour and identity. There is also the challenge that even when the same words or terms are used in a question, these can be understood differently by the person answering the question.
The US census, however, presents an important lesson for future data collection exercises: even when working with quantitative data, we must oppose pressure to airbrush the realities of how identities are lived and experienced. This is particularly true when we face opposition from people who only speak from the standpoint of the heteronormative majority yet seek to police what queer identities are worthy of being counted. Even among queer people, who might enjoy the privilege of being counted, we must ensure we do not pull the ladder up behind us and refuse to settle for an approach to data collection that counts some of us at the expense of all of us.