1. Quality information
Designation of the statistics
On 5 September 2024, Emma Rourke, Deputy National Statistician, wrote to Ed Humpherson, Head of the Office for Statistics Regulation (OSR), to request that the gender identity estimates from Census 2021 are no longer accredited official statistics and are classified as official statistics in development. The change in designation was confirmed by the OSR on 12 September.
This change in designation reflects the innovative nature of the gender identity estimates and the evolving understanding of measuring gender identity, along with the uncertainty associated with these estimates.
All other Census 2021 estimates, including those on sexual orientation, are still designated as accredited official statistics.
Corrected "pansexual" figures
Some responses to the sexual orientation question that should have been coded to "All other sexual orientations" in the detailed classification were incorrectly coded to "Pansexual". This error did not affect estimates using the standard six-category "tickbox" census outputs classification (SO_6A) for sexual orientation but did affect estimates using the nine-category detailed classification (SO_9A) used in the Sexual orientation estimates in the England and Wales Census 2021 statistical bulletin and in our Sexual Orientation (detailed) dataset (TS079).
In November 2023, we published two new datasets containing corrected figures:
We also published a dataset allowing users to see the difference between the correct and original data:
Links to these datasets are provided in the affected products.
The corrected figures show that for England and Wales:
The estimate for "Pansexual" is 48,000 rather than 112,400 as originally published.
The estimate for "All other sexual orientations" is 74,600 rather than 10,200 as originally published.
Figures for all other groups in this nine-category classification are unchanged. Figures for all groups shown in the standard six-category classification are unchanged, including the "All other sexual orientations" category in that classification.
Corrected figures were produced by using write-in responses to recode records initially coded to "Pansexual". The 6,700 records in this group created through the undercoverage adjustment did not have a write-in response and have been assumed to be correctly coded as "All other sexual orientations".
Reliability of estimates
In November 2023, we published a summary of our recent investigations into the quality of census data from the gender identity question and further planned research into this topic. For more information, please see our Quality of Census 2021 gender identity data.
While there are no previous directly comparable estimates available, at the national level, the estimates are considered broadly in line with other data on gender identity. All census estimates contain a degree of statistical uncertainty as described in the Section 6 of our Quality and methodology information (QMI) for Census 2021. This uncertainty will be proportionately higher for relatively small population groups such as those who report a gender identity different to their sex registered at birth. This relative uncertainty will increase if the population group is further broken down by geographic area or other characteristics.
In the November 2023 quality report, we identified patterns in the data that suggested that some respondents may not have interpreted the question as intended, notably those with lower levels of English language proficiency. Analysis of Scotland’s census, where the gender identity question was different, has added weight to this observation. It appears that these respondents in England and Wales were more likely than the general population to say that their gender identity was different to their sex registered at birth, some of whom may have mistakenly given an answer suggesting they were trans. This source of bias would lead to an increase in the estimates of the trans population. While relatively small, it will have greater bearing on the estimates at lower geographic levels, particularly where there are higher levels of non-English speakers.
This is one source of bias among a number. We have previously noted that as a voluntary question on a complex topic, being asked on the census for the first time, we would expect greater uncertainty than we see with other questions in Census 2021. We are unable to quantify the impact of all the sources of uncertainty to give an overall estimate of uncertainty. However, we can say that smaller group breakdowns should not be used as a precise estimate of the trans population and comparisons between areas or groups should be considered with caution, particularly where there are different levels of English language proficiency.
Not answered
As sexual orientation and gender identity were voluntary questions in the census, missing values on census responses were legitimate responses. So, (unlike with mandatory questions) we did not impute values to replace those missing on received responses. This means that these questions include a "not answered" category in standard outputs. Any write-in responses that did not relate to the question were coded as if the question had not been answered. Take care to not confuse between proportions of all people aged 16 years and over falling in a particular category and proportions of all people aged 16 years and over who answered the question falling in that category.
Data collection and coding
As with all census questions, the questions on sexual orientation and gender identity were tested thoroughly as described in our Sexual orientation question development for Census 2021 and Sex and gender identity question development for Census 2021. The final questions collected information on these topics using a combination of tickboxes.
Answers for these questions did not always correspond to well-defined concepts and write-in answers have been coded, both automatically and manually, to categories shown in published tables. While overall estimates for both these topics are consistent with other available sources it is possible that the comparability of individuals’ responses was affected by different interpretations of the questions and that proxy responses for these sensitive topics may be less accurate than for other topics. A proxy response is where someone has answered the census questions on behalf of someone else.
Student adjustment for sexual orientation and gender identity
An adjustment made to improve the estimation of students living at their term-time address will have slightly increased the proportion of “Not answered” and decreased proportions of other categories. Take care when interpreting data for 18- to 22-year-olds or, specifically, full-time students. Other than this, the effect of this adjustment should not materially affect analysis of these topics.
Age and geographical distributions
Census data on sexual orientation and gender identity show age and geographical patterns which may affect relationships between these topics and topics such as education, employment, health and housing. In particular, people who report a sexual orientation other than ‘straight/heterosexual’ and people who report a gender identity different from sex registered at birth are more likely than average to be younger and to live in urban areas. You should be aware of these factors when drawing conclusions from the census estimates for these topics.
Topic summary feedback
Tell us what you think about this publication by answering a few questions.
Nôl i'r tabl cynnwys3. Cite this methodology
Office for National Statistics (ONS), updated 12 September 2024, ONS website, methodology, Sexual orientation and gender identity quality information for Census 2021.