by Laurence Cooley
In December, the British government published its plans for the next census of England and Wales, to be held in 2021. The white paper proposes the addition of three new questions to the census, including one on gender identity and one on sexual orientation. While the census is administered separately in Scotland and Northern Ireland, there is a high degree of co-ordination between the statistical offices in London, Edinburgh and Belfast, and it looks likely that these topics will be introduced on a UK-wide basis; indeed, a bill currently before the Scottish Parliament will add gender and sexuality to the Scottish census.
There is an established history of censuses asking questions about aspects of identity such as ethnicity and race, but gender identity (as distinct from sex) and sexuality are new topics of enquiry, with no country yet having asked about the latter in its national census.[i] Interest in collecting data on these aspects of personal identity is not new, however. Britain’s Office for National Statistics (ONS) launched its Sexual Identity Project in 2006. In the United States, meanwhile, a campaign to “queer the census” was started by LGBT activists as early as 1990.
The 2021 census white paper represents a significant victory for British activists. There is a clearly established need for better data to address inequalities facing sexual minorities. In this post, however, I argue that if privacy concerns surrounding individuals’ disclosure of their gender identity or sexual orientation are not addressed, there is a risk of a significant under-count of the LGBT population, which could play into the hands of those who would attempt to reverse progress towards equality. It is therefore crucial that statistics agencies continue to work with organisations representing LGBT people, to ensure robust solutions to privacy concerns are found and communicated.
Photo credit: PaulSh (cropped)
What’s at stake in counting LGBT identities?
Consultations undertaken by statistical agencies regularly demonstrate that data users, including government departments and service providers, value information about the size and characteristics of sexual minorities. Demand for such data is particularly strong from healthcare providers. LGBT people have worse health outcomes on average than the rest of the population, and in particular are more likely to suffer from poor mental health. Suicide rates for gay and bisexual youth are significantly higher than for their heterosexual counterparts, whereas older LGBT people suffer disproportionately from social isolation and a lack of social support networks.
Moreover, while smaller-scale surveys have been used to estimate the size of the LGBT population, there is significant variation in these estimates. The census has the advantage of being a whole-population count, and so is not susceptible to the same sampling errors. This means that it can be used not only to produce more accurate estimates of the overall size of the LGBT community, but also its geographical concentrations, which is crucial for service planning. Furthermore, the census is used to construct the sample frames for other surveys – and knowing the overall size of the LGBT population means that those surveys can be more accurately weighted to be representative of the population. All of this can help agencies address inequalities faced by LGBT people.
In addition to the practical need for better data on LGBT populations, many people also see symbolic importance in having such a core aspect of their personal identity recognised in official statistics. This was illustrated recently in the US, when the Census Bureau published a list of proposed questions, including sexual orientation and gender identity, as an appendix to its plans for the 2020 census, only to retract them, explaining that these questions had been included in the document inadvertently.
Given concern about the Trump administration reversing progress on sexual rights, activists understandably reacted with shock to this news, arguing that administration was trying to “erase” the LGBT community. Out magazine employed the headline “Trump Administration Omits LGBTQ People from 2020 Census”. Of course, this omission is not a literal one – the census aims to include the entire population regardless of sexuality or gender identity (and indeed, the reporting of the controversy was criticised as “fake news” by some) – but the response was indicative of the symbolic importance of recognition. In the UK, when organisations representing LGBT people welcomed the announcement about the planned inclusion of questions on gender identity and sexuality, they highlighted the practical benefits of collecting this data, but there were also hints of its symbolic importance in the use of language such as “if we’re not counted, we don’t count”.
Given the practical and symbolic importance attached to collecting data on gender identity and sexuality, LGBT organisations in Britain have been overwhelmingly supportive of moves to introduce these questions. In the US, meanwhile, organisations such as the National LGBTQ Task Force were outspoken in their criticism of the decision not to add sexual orientation to the 2020 census, and activists welcomed the introduction of a senate bill by Democrats in July 2018 that would add sexual orientation and gender identity questions by 2030. Nonetheless, there is a tension between this enthusiasm for including such questions and privacy concerns stemming from the same history of discrimination against sexual minorities that provides much of the rationale for collecting this data.
As philosopher Neil McArthur has argued, in the US case, while many LBGT people support the inclusion of such questions, “the idea of having the government collect gender and sexuality data has strong resistance within other corners of the LGBTQ community”. Some of this opposition comes from the libertarian right, where there is long-standing opposition to what are seen as intrusive census questions, but it is not only libertarians who are concerned. Gender studies scholar Jane Ward argues that “under such a conservative presidency, I think people probably would be extra cautious about disclosing” – a caution no doubt heightened by the Trump administration’s desire to add a question on citizenship to the census. Some worry – informed by the experience of Japanese-Americans being interned with the aid of census data during World War II – that this question is intended to help the government locate undocumented immigrants. Such concerns are not limited to the US; England’s termly ‘school census’, which collects data on pupils including their nationality, has been used by the Home Office for immigration enforcement.
The publication of data about the characteristics of local neighbourhood populations – a key benefit of including questions about gender and sexuality on the census – can also raise privacy concerns. Just as it might be useful for planners or healthcare providers to know about geographic concentrations of LGBT people, it is equally possible to imagine such information being used to target or attack those people, as McArthur notes.
Privacy concerns are not restricted to public access to or use of census data, however. When the ONS consulted on including a question about sexual orientation in the 2011 census, it found that there was significant demand for such data, but ruled out adding the question, in large part due to worries about privacy within households. This concern arises because the census form is typically completed by the head of household, which means that LGBT teenagers or young adults still living at home, for instance, would have to be ‘out’ to their parents in order to be counted as such. When Statistics New Zealand tested a question on sexual orientation ahead of its 2018 census, it found that more people refused to answer the question than identified as non-heterosexual. As a result, citing concerns about lack of confidence in the data produced and the sensitivity of the question, the agency opted not to include it in the final census form.
Technology might offer a potential solution to such sensitivities about privacy. The ONS anticipates that 75 per cent of people will complete the census online in 2021, and the white paper setting out plans to include questions on gender identity and sexual orientation suggests that should they wish, individuals will be able to request access to their own, private census form without having to inform the head of household. Moreover, the questions on gender and sexuality will only apply to those aged 16 and over and answering them will not be compulsory.
Risk of an under-count
Even if a proportion of respondents will always prefer not to disclose their sexuality even on a confidential form, surely having data – however imperfect – on the size of the LGBT population from the census is better than not trying to collect that data in the first place? While I agree with this sentiment in principle, I do think that there are potential risks associated with an under-count – particularly if it is large.
One danger is of a backlash against the inclusion of questions about gender identity and sexuality in the census in the first place. Some commentators have already argued that these questions are being driven by ‘political correctness’ rather than statistical need. While it might seem easy to dismiss these arguments by pointing to the well-established demand for the data, this backlash is arguably more about ‘anti-gender’ ideology than rational concern for evidence. Populist politician Maxime Bernier, for example, reacted to the news that the topics to be covered by a pilot for the next Canadian census would include gender by tweeting about the “gender-obsessed Trudeau government”.
An under-count might also provide ammunition for critics of efforts to improve LGBT representation. The Australian experience is instructive here, and demonstrates that employing technology alone is not a silver bullet. The 2016 census recorded only 1,260 people who considered themselves sex and/or gender diverse, but the Australian Bureau of Statistics (ABS) notes that this “is not considered to be an accurate count, due to limitations around the special procedures and willingness or opportunity to report as sex and/or gender diverse”. To identify as other than male or female, respondents had to request access to a special online form, with the standard form distributed to households retaining binary response options. This will clearly have limited the number of people counted as ‘gender diverse’, but nonetheless the 1,260 figure has been used by some to argue that transgender issues are receiving disproportionate attention. An article in the Australian edition of Spectator magazine about a television station’s decision to feature an all-transgender panel, for example, noted the ABS’s caveat about the likely under-count, but nonetheless mockingly quoted the figure: “Such vast numbers! One thousand, two hundred and sixty out of a population of 24.3 million”.
Similarly, the Belfast-based News Letter newspaper recently used survey statistics published by the ONS to argue that gay and bisexual people are over-represented in the Police Service of Northern Ireland – despite the fact that those statistics are considered experimental and, given that they are survey-based, might well under-estimate the proportion of the population that identifies as LGBT, as organisations such as Stonewall have pointed out. This fits a broader pattern, identified by Ward, of conservative voices using surveys to claim that “we’re far too worried about protecting LGBTQ people given how small their population is”. While the News Letter article was rightly criticised on social media, one can imagine such arguments being harder to push back against if they cite seemingly more authoritative census figures.
So, while progress is clearly being made towards the inclusion of gender and sexuality questions in censuses – American setbacks aside – I think that it is too early to claim victory in the battle for more accurate statistics on LGBT populations. Inclusion of these questions is a major milestone, but there is still work to be done to avoid the dangers of a significant under-count. Here, we could learn from the census experience of ethnic and racial minorities. When advocacy groups succeeded in getting the US census to count the Hispanic population starting in 1980, community leaders subsequently worked with the Census Bureau to inform people about the importance of being counted as Hispanic and to build trust in the census process. As and when questions on gender identity and sexual orientation are added to censuses, statistical agencies need to continue to engage with organisations representing LGBT people to ensure that privacy concerns are fully addressed, and that solutions to these privacy issues are properly communicated to LGBT communities, who can then have greater confidence in being counted.
Laurence Cooley is a recent ESRC Future Research Leader, and teaches in the School of Government and Society at the University of Birmingham. He is also Visiting Research Fellow in the School of Natural and Built Environment at Queen’s University Belfast. His current research is about the relationship between political representation and the politics of the census in deeply divided societies.
[i] In this post, I focus primarily on the reasons for and challenges of collecting this data, but I acknowledge that other questions might be posed here. Is the decision to maintain binary responses to the question on sex a missed opportunity to address the lack of robust data about the size of the intersex population, for example? How does asking about sexual orientation, as opposed to sexual identity, attraction or behaviour, shape who is counted? And do the response options offered sufficiently reflect the potential fluidity of gender identity and sexuality? These questions deserve further attention in future work on the census, gender and sexuality.