Ed Humpherson to Mary Gregory, Delphine Robineau, Richard Elliott and Stephanie Howarth: Presentation and reporting of census sexual orientation statistics
Dear Mary, Delphine, Richard, and Stephanie
Presentation and reporting of census sexual orientation statistics
I am writing following concerns raised by the LGB Alliance and The Lesbian Project about the use of the term ‘LGB+’ in Census 2021 sexual orientation statistics. Thank you to you and your teams for your engagement on this case and providing us with the relevant information required.
Although the concerns focused on the Office for National Statistics (ONS)’s statistics for England and Wales, we considered the approaches taken across the UK to the aggregation, definition, and communication of sexual orientation data collected through the census.
Our findings
Our findings support the central concern raised with us that statistics for England and Wales have been misreported by third parties. The examples provided show instances where census sexual orientation statistics produced by ONS have been described by third parties as ‘LGBT+’ or ‘LGBTQ+’ statistics, potentially giving users the incorrect understanding that the figures include people on the basis of gender identity. There are also examples of the ONS statistics being reported appropriately. We are not aware of misreporting in Northern Ireland and Scotland and therefore have not found evidence that misunderstanding is widespread across the UK. We note that census producers were not aware of any widespread misreporting before we shared these concerns. The classification ‘LGB+’ was developed for Census 2021, the first census to produce data on sexual orientation. ‘LGB+’ is used to report aggregated sexual orientation statistics by the census offices as the data represents how questions were asked in the census specific to the options of ‘gay or lesbian’, ‘bisexual’, or ‘other sexual orientation’. ONS confirmed that there was no alternative abbreviation or term in use that would cover the same range of responses as ‘LGB+’.
Each of the census offices and the Welsh Government generally provide appropriate definitions alongside its census statistics, explaining which responses are included within ‘LGB+’. However, this guidance does not explicitly address the potential risk for the aggregated sexual orientation statistics to be confused or conflated with terminology that encompasses gender identity.
We also identified some differences in the terminology and presentation used across the UK. While ‘LGB+’ is used in reporting census sexual orientation statistics, producers have also used different terms in response to additional user requests or feedback. For example, the Northern Ireland Statistics and Research Agency uses the term ‘LGBO’ in a commissioned output using the same language as the requestor. Welsh Government uses the term ‘another sexual orientation’ to describe the group ‘other sexual orientation’ based on feedback from Wales-based equality communities. These approaches are not incorrect, and the Office for Statistics Regulation (OSR) is not seeking to prescribe a particular statistical term or definition. However, differences in presenting broadly comparable information may create confusion if the differences and the reasons for them are not clear.
Scotland’s 2022 Census included additional questions on trans status or history, with NRS presenting these statistics alongside sexual orientation statistics in the same bulletin. NRS has separate sections to report estimates on the topics and provide definitions. We welcome NRS’s subsequent removal of the standalone definition of the ‘+’ symbol from its published statistics, and the further explanation NRS has provided on what the ‘+’ symbol means in each of the different abbreviations. This further clarifies meaning and supports appropriate interpretation.
OSR considers when the term ‘LGB+’ is used to report statistics, particularly the ‘+’ symbol may carry a risk of misinterpretation. The meaning of ‘+’ may be understood differently, given the wider societal use of abbreviations such as ‘LGBT+’ and ‘LGBTQ+’. The misreporting identified through this casework demonstrates instances of inaccurate reporting but does not establish whether the terminology contributed to them. Producers should assess whether LGB+ is the most suitable term and consider clearer explanations or changes to presentation to reduce the risk of misinterpretation.
Harmonisation and guidance
We have identified a potential gap in existing harmonisation guidance on this topic. Current Government Statistical Service (GSS) guidance focuses primarily on the design and collection of the sexual orientation question rather than how the resulting statistics should be aggregated, labelled and communicated.
Given the sensitive nature of these statistics and the potential for misinterpretation, we recommend that the GSS Harmonisation team develops additional guidance to promote consistency, recognise legitimate differences, and help users understand and report these statistics accurately.
Our expectations for producers
Alongside this work, we expect census statistics producers to take proportionate steps to reduce the risk of further misinterpretation. In particular, we ask you to:
- Review and assess the suitability of the term ‘LGB+’ to report 2021/2022 sexual orientation statistics, considering the risk of misinterpretation and ensuring that any changes to terminology remain inclusive.
- Strengthen the prominence of current information explaining the meaning and coverage of aggregated sexual orientation statistics presented alongside Census 2021 and 2022 statistics. This should include providing advice to users on appropriate and inappropriate ways to report the statistics, including the terminology they should use to clearly distinguish the representativeness of these estimates from the broader LGBT+ and LGBTQ+ population for example, adding a prominent notice or banner at the top of statistical bulletins and any supporting data tables.
- Follow relevant harmonised standards where these are available and, where a different approach is used, explain the reasons for this clearly and transparently, including any implications for user interpretation or comparison of the statistics.
- Challenge the examples of misreporting of ONS statistics identified through this casework and take appropriate corrective action, including engaging with the organisations concerned or issuing clarification of corrective notices where appropriate.
- Establish an approach for identifying and challenging any future instances of misreporting to support accurate use.
- Use the evidence from this casework and the census topic consultations to inform the development and testing of Census 2031 questions, categories and output terminology. Testing for Census 2031 should assess understanding of published terminology separately from understanding of the question itself.
Next steps
Overall, we consider that the action to prevent future misreporting should be proportionate. Clearer definitions and terms, more prominent advice for secondary use, and stronger harmonisation guidance should help ensure that census sexual orientation statistics are communicated accurately and consistently. We also note that the findings of this casework may be relevant to other statistics producers that may collect and produce data on this topic. We request that you respond publicly with an update on the actions you have taken or intend to take in response to our recommendations before the end of January 2027.
I am grateful for the constructive engagement from your teams throughout this case.
I am copying this letter to Fiona Dawe, GSS Harmonisation team; Jane Naylor, Director, National Statistician’s Office; Alastair McAlpine, Chief Statistician, Scottish Government; Philip Wales, Chief Executive, Northern Ireland and Statistics Research Agency; and the Statistics Head of Profession Office at ONS.
Yours sincerely,
Ed Humpherson
Director General for OSR