3. Expectations for further work
3.1 Throughout our review, it has been clear that the ONS team leading the quality assurance and additional exploration of the data is committed to understanding the quality of the data and how this can best be communicated to users. The team has sought innovative ways to interrogate the data, including subgroup analysis and data linkage. It has also ensured to keep NRS updated on its findings as it goes along and NRS told us that ONS has been incredibly helpful.
3.2 We expect ONS’s research on gender identity to seek to understand to what extent, if any, has a misunderstanding of the gender identity question in the 2021 England and Wales Census led to a significant misestimation of the transgender population. When the target group is small, the effect of false positives and negatives on the data will be amplified.
3.3 ONS should consider the testing and operationalisation of the question as well as the production and analysis of the results. If there is evidence that there has been a significant misestimation of the transgender population to a degree that might affect usage, the research should set out both the implications for the use of the results from the Census and the use of the GSS harmonised question in future data collection.
3.4 Beyond the use of the 2021 results, the Future of Population and Migration Statistics consultation document highlights issues with lack of coverage of administrative data on gender identity. It is therefore crucial that ONS understands the performance of this question through the additional research and determines how it will look to collect these data in the future. This should be considered alongside its approach for the development of UK wide estimates on gender identity.
3.5 We expect ONS to take on board the following recommendations as it works on developing its research.
3.6 Recommendations to support further use of the data:
- ONS should publish a statement within four weeks of this interim report that sets out a timeline for its ongoing research, covering what it is analysing and what it plans to publish by when. This statement must be open about the potential quality issues it is investigating, even if it cannot quantify the scale of them. This is important both for the use of published estimates, but also the use and further development of the harmonised question.
- ONS should make data available to understand subgroups of the reported transgender population. For example, it should support analysis of differences in responses for those who provided an explicit and unambiguous write-in response indicating that they were transgender compared with those who provided a possibly ambiguous or tick-box only response.
- ONS should consider any potential misunderstanding of the question or mode effects on the data as part of its analysis into the agreement rates for the gender identity question in the Census and Census Quality Survey. In particular, it should publish analysis of group differences to inform future use of the data and question.
- As part of the Future of Population and Migration Statistics consultation and programme, the UK harmonisation group should set out its early thinking on gender identity estimates for Great Britain and publish its development plan for these statistics.
3.7 Recommendations to support further use of the question:
- As part of the ongoing development of the harmonised standard for gender identity, should the question remain the same, ONS should carry out further testing of the question. This should be considered as part of the outcome for the Future of Population and Migration Statistics consultation, in determining future production of these statistics in an administrative data based framework.
- ONS should publish more details on the feedback it received in the Census rehearsal, including any insight about those with lower English language proficiency.
- Once NRS has published its data on gender identity from the 2022 Scotland Census, planned for 2024, ONS and NRS should carry out a lessons learned exercise comparing the performance of the two questions and use this to inform future developments and use of the questions.