Data matters. Good data matters more. At a cost of £1bn, it is hard to think of a larger or more important data-collecting exercise than the census. Its seriousness is reflected in the £1,000 maximum fine and criminal record for those who provide false information or refuse to answer questions.
The Office for National Statistics (ONS), which carries out this once-in-a-decade exercise, describes it as “vital to the government and many other public sector organisations”; it gives them the information they need to develop policy, plan and run schools, health services, roads and libraries, and to decide how to allocate funds. So when the accuracy of that information is called into question, it really matters. And should worry us all.
Last week, census data was downgraded in status from “accredited official statistics” to “official statistics in development” – the first time this had happened since the census began in 1801. The 2021 survey asked two new questions: one about sexuality, the other about gender identity. Both were voluntary. Neither have been entirely without controversy: in November 2023 the ONS corrected an error relating to those identifying as “pansexual”. The number had been overestimated by nearly 65,000. But the question that aimed to record the transgender population of England and Wales has been most problematic. This asked: “Is the gender you identify with the same as your sex registered at birth?” According to the ONS, 262,000 people – around 0.5 per cent of the population – answered “no”.
A review undertaken by the Office for Statistics Regulation has now found that those who do not speak fluent English may have misunderstood the question, mistakenly answering that they considered themselves trans. A week earlier, the ONS had asked the regulator to reclassify the data, but up to then had been highly defensive of it.
When results for the new questions were first released in January 2023, some argued they weren’t credible. The Oxford sociologist Professor Michael Biggs highlighted idiosyncrasies. According to the census, the London boroughs of Brent and Newham had a higher proportion of transgender people than anywhere else – 1.5 and 1.3 per cent, respectively. Brighton and Hove, widely seen as Britain’s most LGBTQ-friendly city, ranked 20th with 1 per cent.
At Biggs’ request, the ONS cross-tabulated the gender identity question with a question on English proficiency. And sure enough, it confirmed that a surprisingly high proportion of people who had been recorded as identifying as trans did not speak English (or Welsh in Wales) as a first language. While only 10 per cent of people said they did not speak English as their main language on the census as a whole, this same group appeared to make up 29 per cent of the total number of transgender people. Those who spoke English “not well” or “not well at all” were more than five times as likely to be counted as transgender, compared with those whose main language was English (or Welsh).
To Biggs, it seemed abundantly clear something was wrong. Yet the ONS dug in. Defending its methodology and accuracy in November 2023, it argued that “while these patterns [whereby areas with large ethnic minority populations had a higher rate of trans responses] may be unexpected, this does not mean they are necessarily wrong”. And even though it admitted the follow-up Census Quality Survey (where a small sample of the population is phoned to double-check their answers) found much lower agreement rates for people who had answered they were trans on the census than for those who didn’t, the ONS argued people could be censoring answers due to the “risk of others in a household overhearing”. The words “straws” and “clutching” come to mind.
What happened? Biggs has suggested the ONS had allowed itself to be lobbied by LGBTQ pressure groups. In 2018, for example, the ONS evaluated the question “via community testing at LGBT History Month events”. When it later tested its wording on 300,000 households, it found “some respondents found this question difficult to answer”. The question got a risk rating of “medium” for its “potential for impact on data quality”. While the statistics regulator has rejected allegations that the “ONS has been captured by interest groups”, it said: “We consider it regrettable that ONS’s defensiveness has created an impression of bias to some external observers.”
So, a once-in-a-decade opportunity to gain an accurate picture of the trans community, and potentially better address its needs, has been squandered. Public policy must be determined by good data. Yet the same unclear question on gender identity has been used in the annual GP Patient Survey since 2021. When BMJ Open published an article on the healthcare needs of the transgender population using this survey, Biggs tried to issue a warning. BMJ Open refused to publish it. Biggs alleged in a recent article for the Spectator that he had obtained emails between BMJ staff noting he was “known for being transphobic”.
When government bodies, academia and medicine are more concerned with where criticism is coming from, rather than its veracity, there’s a problem. And when ideology trumps evidence, especially in healthcare, it is dangerous. Questions framed through fear of causing offence or to placate certain minority groups render the answers worthless, and a waste of increasingly scarce public money.
[See also: How will the gender care report affect politics?]
