Inclusive Data Principle 6
Broaden the range of methods that are routinely used and create new approaches to understanding experiences across the population of the UK.
We will undertake research using innovative methods best suited to the research question and prospective participants, to understand more about the lived experiences of several groups under-represented in UK data and evidence.
The ONS Centre for Equalities and Inclusion will facilitate collaboration across the planned initiatives to understand more about the lived experiences of several groups under-represented in UK data and evidence, identifying and sharing good practice.
The development of the Integrated Data Service (IDS) by ONS as a tool to enable improved data linkage and intersectional analysis across government and wider organisations will play a key role in improving UK data infrastructure. ONS is developing the IDS to significantly improve and increase access to, and use of, data from across the UK, by analysts in UK Government Departments, Devolved Administrations, the National Health Service and Local Authorities. This service is designed to enable quick and convenient access to researchers, while protecting confidentiality of data subjects at all times, using extensive technical and operational controls, and robust and transparent governance.
The new independent National Statistician’s advisory committee on inclusive data will advise the National Statistician on approaches in undertaking qualitative research in order to better understand the lived experiences of these groups of people in the UK.
Current and planned work
- Feasibility work on how new and big data sources can be used to fill data gaps will be undertaken, led by the ONS Data Science Campus in collaboration with international members of the United Nations Committee of Experts on Big Data and Data Science (UNCEBD). We will review other countries’ experience of using big data to fill inclusive data gaps, working with the UNCEBD and relevant ONS colleagues.
- ONS is developing a non-official sources protocol to facilitate quality assurance of non-official sources and maximise their use while minimising risks. The protocol for quantitative sources will be implemented in 2022, with the development of a protocol for qualitative sources in 2023.
- ONS is researching the coverage of specific administrative datasets to better understand how certain groups within the population are represented. Qualitative research methods are also being developed to give us a greater insight into any inclusivity issues for such data sources.
- During 2022, ONS is undertaking qualitative research to examine the lived experiences of groups who are currently under-represented in UK data and evidence including: disabled adults’ experiences accessing and engaging with activities, goods and services across the UK; the school experiences of children with special educational needs and disabilities in England; and the lived experiences, priorities and needs of Gypsy and Traveller communities.
- The Cabinet Office’s Disability Unit (DU) will undertake a systematic review of evidence in relation to the lived experiences of disabled people in the UK, to take stock of relevant qualitative research and identify key gaps in understanding and priorities for further work by the end of March 2022.
- ONS is continuing to investigate improved methods for data linkage, as part of the Joined up data in Government Review. New tools and techniques are being investigated to both improve linkage quality and enable better measures of linkage uncertainty and bias. These will be an important part of any data analysis which uses linked data, especially for population subgroups.
- As part of the continued drive to explore Data Science methods, ONS will be examining potential biases in machine learning algorithms and when they can result in outcomes which may not be fully inclusive. These techniques are heavily dependent on the representativeness of the training data, perpetuating any biases through to the model. Work is underway to explore de-biasing methods, as well as how Machine Learning algorithms used in production are maintained. This will include managing the risk of model bias, and thus lack of inclusivity, growing over time due to model drift.
- Data are rarely complete, especially when linking sources, and the methods used to deal with missing data can have an impact on inclusivity. ONS will be examining strategies for imputation in the context of linked administrative data and evaluating the impact of missingness which accumulates due to linkage errors or lack of coverage. This will extend the work already completed to explore the imputation of administrative based income linked to Census data, which found important challenges in the representation of the lower income distribution.
- As part of the transformation of the population social statistics system, ONS will develop longitudinal assets. These will enable the flagging of disadvantaged groups and of transitory states of interest to support subsequent analysis. The integrated statistics system will be designed and built in 2022.
- Department for Educationwill continue its Longitudinal Study of Young People in England, following a cohort of young people aged 13/14 in 2013 through the final years of compulsory education, and onto other forms of education, training, employment and other activities, collecting information about career paths and the factors affecting them and a range of characteristic information. In 2021/22 the questionnaire will be developed, and fieldwork carried out, with data available in 2022/23.
- Department for Education’s Education and Outcomes Panel Studies (EOPS) programme will deliver two new cohort studies following children from 9 months to 5 years of age (Children of the 2020s) and from early in their primary education (Years 1 or 2) to the end of primary school (Year 6) (Pupils of the 2020s). Pupils of the 2020s will be procured and set up in 2022, with fieldwork and delivery of the first wave of data in 2022/23.It will use Free School Meals eligibility criteria from the National Pupil Database to oversample disadvantaged pupils, and better understand the relationship between attainment, disadvantage and a range of personal and household characteristics.
Data producers will ensure that data linkage projects are routed through the appropriate governance and approvals mechanisms, for example:
- Home Office (HO) will ensure that all work with ONS on data linkage is approved through appropriate programme governance with senior representation from both HO and ONS.
- ONS will ensure that all transformed statistical production is reviewed and assured through formal channels (Census Research Assurance Group, Methodological Assurance Review Panel and Longitudinal Scientific Advisory Panel) and that all linkage studies have approval from the National Statistician’s Data Ethics Advisory Committee. The transformation of our surveys will likewise be reviewed and assured through appropriate forums.
- ONS will ensure that all datasets used in analysis of the 2021 Census for England and Wales will have been approved by the relevant ethical boards prior to the start of analytical work.