Kate Barker and Diane Stoianov to Ed Humpherson: Presentation and reporting of census sexual orientation statistics
Dear Ed,
As members of the lesbian interveners, we provided written submissions to the Supreme
Court on the winning side in For Women Scotland Ltd v The Scottish Ministers.
We are now writing to ask the OSR to review the ONS’s introduction and continued use of
the label “LGB+”. This label was introduced in the 2021 Census to describe the “minority
sexual orientation” population, but it does not meet basic standards of statistical production
and is being widely misunderstood.
Since the Census results were published, public bodies and other competent users of these
accredited official statistics have repeatedly mistaken the LGB+ sexual orientation statistic
for an umbrella LGBT+ statistic, erroneously conflating at least two distinct populations.
1. Our analysis
In its Census 2021 output, the ONS introduced the apparently novel term LGB+ as its
headline category to denote people with a “minority sexual orientation”. Its reasoning seems
to have been to include people who were not heterosexual, but identified with an alternative
term to lesbian, gay or bisexual.
However, the coalition umbrella term LGBT+, which includes people with transgender and
other identities unrelated to the sexual orientation question, was already well-established in
ordinary English usage.
BBC Online, for example, maintains a dedicated LGBT topic page, categorises news stories
with the LGBT tag, and contains thousands of references to the terms LGBT and LGBT+.
The coalition has, somewhat notoriously, expanded over time, with ‘+’ conventionally being
added at any point to indicate the remaining members:
LGB
LGBT
LGBTQ
LGBTQI
LGBTQIA
LGBTQIA+
A user encountering the LGB+ label has therefore every possible reason to interpret it as
including the familiar identities it is normally associated with.
The ONS, however, appears to expect the user to know that – in just this case – the ‘+’
actually means additional sexual orientations not normally associated with the initials (the
Census guidance suggested “asexual” and “pansexual”), but crucially not the gender
identities and other identities it is normally associated with.
Furthermore, an elementary reason to expect the term to be confusing is simply that LGB+
differs by exactly one character from LGBT+. This makes straightforward misreading
inevitable, even for those users who have understood the novel meaning of the label.
This appears to be an unprecedented mistake. It is hard to comprehend how the ONS did
not foresee that the term LGB+ would be predictably and inevitably confused with the term
LGBT+, causing widespread misinterpretation of the Census sexual orientation statistics.
2. Examples
Department of Health and Social Care (2023)
In their Equality impact assessment – minimum service levels in event of strike action:
ambulance services the DHSC says that the Census found “around 3.2% identifying as
LGBT+ (lesbian, gay, bisexual and transgender)”.
However, the 3.2% figure is precisely the LGB+ sexual orientation value, which does not
include gender identities or transgender status. The DHSC misunderstood the category, and
incorrectly added transgender identities to the statistic.
Stonewall (2023)
The UK’s leading LGBTQ charity states: “England and Wales’ 1.5 million LGBTQ+ people,
and their families, deserve better.” The article makes over ten references to LGBTQ+, but
sizes this population using a value that can only have come from the Census 1.5 million
LGB+ sexual orientation statistic.
Stonewall statement on the UK Government’s failure to ban conversion practices | Stonewall UK
Leicestershire County Council (2023)
Leicestershire’s statutory Joint Strategic Needs Assessment about alcohol misuse states
that 14,292 people were in “LGBT+ groups”. It goes on to say that the Census sexual
orientation data “corresponds to the LGBT+ population”. In both cases, LGB+ is mistaken for
LGBT+.
LEICESTERSHIRE COUNTY & RUTLAND NHS PRIMARY CARE TRUST
Science Museum Group (2023)
Its 2022/23 Diversity Data Report says “We do, however, have high levels of LGBTQ+
representation at 22.7% of employees compared with 3.2% who identified with an LGBTQ+
category in the 2021 census.” There was no LGBTQ+ category; this is the LGB+ value.
SMG Workforce Demographics July 2022
Pink News (2023)
In the story Census reveals which parts of England and Wales have most LGBTQ+ people,
the headline and image caption refer to Kemptown’s LGBTQ+ population, while the article
itself says the figures concern the 3.2% LGB+ population and explicitly says its percentages
are based on the LGB+ figure.
Despite being a specialist LGBTQ news outlet, and even though the author must have had
the source material in front of them, the article slips indiscriminately between LGB+, LGBT+
and LGBTQ+.
2021 census map: LGBTQ population in England, Wales revealed
Home Office (2024)
The Draft domestic homicide review statutory guidance reports that 3% of victims were
“recorded as being LGBTQ+”, then says that “the prevalence of LGB+ victims in domestic
homicides matches the figure of LGB+ individuals recorded in the general population data”.
The guidance moves between LGB+, LGBT and LGBTQ+ formulations in the same
discussion.
Draft domestic homicide review statutory guidance (accessible) – GOV.UK
British Medical Association (2024)
The BMA article Why we need LGBTQ+ teaching in medical education notes people
identifying as LGBTQ+ constitute a significant proportion of the population – “3.2% according
to the 2021 census”. It explicitly and mistakenly discusses transgender people as part of this
population.
Buckinghamshire Council (2024)
In its Workforce equalities profile 2023 to 2024, Buckinghamshire Council states that nearly
10,000 residents “identified as LGBT+ as part of the Census 2021” and contrasts that figure
against the national 3.2% LGB+ sexual orientation statistic.
Sexual orientation | Buckinghamshire Council
Southwark Council (2024)
Similarly, Southwark Council reports in its Equality Impact and Needs Analysis that 8.08% of
residents “identified as LGBT+” in Census 2021, but cites the LGB+ sexual orientation
statistic.
Regulator of Social Housing (2024)
In its Consumer standards – Equality impact assessment the RSH says “3 per cent of those
living in social housing identify as LGBT+” and immediately afterwards that “0.8 per cent of
social housing residents do not identify with the sex registered at birth”. Both figures are
footnoted to the Census.
ONS publishes two separate Census datasets: sexual orientation by tenure, and gender
identity by tenure. RSH has taken the LGB+ figure and described it as LGBT+, despite
reporting the gender identity statistic on the next line.
Annex 6: Consumer standards – Equality impact assessment – GOV.UK
LGBT Foundation (2025)
On the day the Census results appeared in January 2023, LGBT Foundation correctly
reported that 1.5 million people (3.2%) “identified with an LGB+ orientation”. Then, in a major
2025 policy document, LGBTQ+ Health Mission for a Better Britain the foundation says: “The
2021 Census told us for the first time that at least 1.5 million LGBTQ+ people reside in
England and Wales…” This is a formal health policy document from an organisation whose
entire field is LGBTQ issues, but it instinctively converts “1.5 million LGB+” into “1.5 million
LGBTQ+”.
Brunswick Centre & Pennine Domestic Abuse Partnership (2025)
A report Understanding LGBT+ experiences of domestic abuse in Kirklees, produced
through a partnership involving the Brunswick Centre, PDAP and Kirklees Council, gives the
Census sexual-orientation count of 1,536,557 and explicitly says it is the Census definition of
LGB+, “which for the purposes of this report, will be treated as LGBT+.” It labels the LGB+
figure in its infographic as people who “identify as being LGBT+”.
Understanding-LGBT-experiences-of-domestic-abuse-in-Kirklees-2025.pdf
European Sociological Review, Oxford University Press (2026)
The peer-reviewed paper Educational outcomes of UK LGB and heterosexual adolescents:
exploring gendered differences and mechanisms states: “According to the latest data of UK
CENSUS (2021) on sexual orientation and gender identity, around the 3.2 per cent of the
population aged 16 years identified themselves as LGBTQIA+.” It mentions gender identity,
and cites the ONS Census sexual orientation article.
Despite undergoing professional academic authorship, peer-review and editorial production in a leading sociology journal, the 3.2% LGB+ figure becomes LGBTQIA+
Journal of Surgical Education, APDS (2026)
The peer-reviewed paper Diversity in Surgery: Ethnic, Gender, and LGBTQ+
Representations Among United Kingdom Surgical Residents and Consultants repeatedly
describes the Census 3.2% LGB+ sexual-orientation estimate as an LGBTQ+ population
prevalence and uses that value as the reference population parameter for inferential
comparisons reported as LGBTQ+ representation.
Despite undergoing professional academic authorship, peer-review and editorial production,
the 3.2% LGB+ figure becomes LGBTQ+.
There are innumerable further examples of this confusion in mainstream local and national
news media.
3. The Code of Practice
When the Census results were published, the OSR Code of Practice for Statistics version
2.1 required statistics to be presented in a “clear, unambiguous way”. It required their
meaning to be explained in a way that was not materially misleading (V3.2) and it required
the Head of Profession to challenge inappropriate use of statistics and “reflect upon how
further misuse can be prevented” (T2.5).
The current Code, version 3.0, is even more explicit.
● Standard 9, “Be clear”, requires producers to present statistics so that they support
appropriate interpretation.
● Practice 9.5 requires producers to “Aid understanding by highlighting potential
misinterpretations.” and to “Clarify the meaning of the statistics when they are used
inappropriately.”
● Practices 9.1 and 9.2 require clear communication to support appropriate
understanding.
Numerous examples, easily found, show beyond doubt that widespread misinterpretation of
the headline LGB+ statistic has occurred and continues to occur. Examples can be found
from central government, local government, regulators, general charities, specialist LGBT
charities, medical, accountancy and engineering professions, and local, national and
international media.
It therefore does not meet the requirements of the Code.
4. The use of ‘+’ is statistically indefensible
The reason for this mistake will be apparent to anyone familiar with ONS’s work on sex and
gender during the last decade. As with its other engagements in this area, normal scientific
practice and standards of clarity were abandoned.
A plus sign is perfectly clear where it has a mathematically determinate meaning: 65+,
£100,000+, five or more employees. In those cases, ‘+’ means a precise inequality: x ≥ n.
The category is open-ended numerically, but the membership rule is closed and objective.
That is not what LGB+ does. Here, the plus sign is not an operator over an ordered variable,
but means something like “and other sexual orientations that belong with these ones”. But
that immediately raises questions that ordinary statistical practice is supposed to settle
rather than create:
● Which other orientations?
● By what criterion are they included?
● Are identities such as asexual included? Pansexual? Queer?
● Does the set change over time, when new labels enter common usage?
● Is membership determined by identity, behaviour, attraction, or simply the Census
response coding?
Even if there were no similarity between the LGB+ and LGBT+ labels, there is no other
category in official statistics whose membership is left open to the user’s imagination.
An open-ended category is not a category. A category should have definable inclusion
criteria, have a stable meaning, permit intercoder reliability and survive ordinary-language
interpretation. “Plus” does none of these. What does it include?
● Polyamory?
● Kink identities?
● Gender identities?
● Romantic orientations?
● Paraphilias and fetishes?
Once a classification system adopts an open-ended “plus” category, it has abandoned
taxonomic discipline.
A Census should not adopt categories whose semantic boundaries are politically contested
and rapidly mutating. Bizarrely, the statistical category itself was finite and operationally
defined, yet ONS gave it an open-ended, activist-inspired public label.
The use of “plus” is statistically absurd. Considered alongside ONS’s unprecedented public
failures on both sex and gender identity in the 2021 Census, it is residual evidence of
2020-era political activism within the organisation which it failed to understand or recognise.
5. The law
Section 12 of the Equality Act 2010 provides that sexual orientation means a person’s sexual
orientation towards:
(a) persons of the same sex,
(b) persons of the opposite sex, or
(c) persons of either sex
The Supreme Court has confirmed that Equality Act concepts must retain coherent,
biologically grounded meanings throughout the statutory framework.
The sexual orientation statistic is a headline equalities benchmark. ONS classifications that
operationalise sexual orientation through unstable, identity-based or activist-derived
nomenclature are inconsistent with the clarified legal landscape in the UK and may be open
to judicial review.
6. Questions for ONS
We would be grateful if ONS could answer the following questions.
- What evidence does ONS have that users understand the distinction between LGB+
and LGBT+? - Was the label LGB+ tested for comprehension? (This is distinct from the extensive
cognitive, quantitative and UX testing ONS says it conducted on the sexual
orientation question itself.) - Has ONS taken any action in response to documented instances of its LGB+ statistic
being misunderstood or misreported as an LGBT+ statistic? - Will ONS review whether LGB+ should continue to be used as the headline statistical
label for sexual orientation?
Our concern is not simply that the ONS has made a mistake, but that the mistake is
elementary and obvious. It would not have been made in any other area of statistical
production. It is obvious that LGB+ is a confusing, ambiguous, absurd label that inverts the
norms of statistical production. It is obvious that it would be widely misused and
misunderstood.
The 2021 Census sexual orientation results, as presented, do not meet basic standards of
accredited official statistics. We think that ONS should cease use of the term LGB+ in
statistical outputs, and revert to producing clearly defined, bounded, and conceptually
coherent categories compatible with the law.
It is now past time for the ONS to undertake a full, transparent and honest review into how
this vital institution went so wrong in the area of sex, gender and sexual orientation.
We are publishing this letter, and we request that it and any response is also published on
the UKSA website in a way that makes it findable to users of the sexual orientation data.
Yours sincerely,
Kate Barker
CEO, LGB Alliance
Diane Stoianov
Research Lead, The Lesbian Project