Communicating the quality of ethnicity data
We really enjoyed our session at the 11th European Conference on Quality in Official Statistics (Q2024) conference in Estoril, Portugal (Greater Lisbon region). The Equality Data and Analysis Division in the Cabinet Office delivered a presentation on how we communicate the quality of ethnicity data. Here is a summary of our thinking.
At the heart of what the Equality Data and Analysis Division (EDAD) is trying to do is steer a way through three main issues:
- Quantity of data
- Data quality and interpretation
- Communicating data quality
Quantity of data
There is a lot of data in the UK about people’s ethnicity. This includes administrative data on:
- health
- education
- crime and the criminal justice system
Survey data exists for topics such as:
- labour market outcomes
- housing
- culture and heritage
- transport
The Equality Hub’s Ethnicity facts and figures website showcases a lot of data about ethnicity for these topic areas.
Data quality and interpretation
There is a lot of ethnicity data produced by the government, and it is collected in different ways, using different classifications. This leads to many different quality issues. Some of these issues can make the data hard to use. Users of the data need to be aware of a range of potential pitfalls. It might make the data unfit for purpose depending on the needs of the user.
Ethnicity as a sensitive topic
‘Ethnicity’ is a relatively sensitive and high-profile topic. Data needs careful interpretation to avoid being misleading.
Our role
Equality Hub analysts have a role to play in facilitating responsible and informed discussion around the different outcomes for people from different ethnic groups. This role enables us to support policy interventions and to stimulate debate. To be able to do this effectively, it is important that our analytical work is trusted. And to work with UK government departments to improve the quality of their data, it is important that our expertise is respected.
However, we cannot simply expect to be trusted and respected. We have to behave in ways that are regarded as trustworthy, and to demonstrate our professionalism. To do this we have sought to actively manage our reputation.
For example, we engage extensively with users, so that we have an identity and a personality. One way we do this is by communicating in our own names and giving talks to users from both inside and outside government.
We also describe the statistical information we present in neutral and objective terms. We make sure it adds value to the underlying data.
Communicating data quality
These approaches are pretty standard for government analysts. What we do differently is in our strategic approach to describing the quality of ethnicity data. We have adopted an approach that takes advantage of three different ways to communicate ethnicity data quality:
- using a set of standards aimed mainly at producers of ethnicity statistics
- a series of technical reports
- highly accessible blog posts
Standards for Ethnicity Data
Our innovative and well-received Standards for Ethnicity Data extract significant features of the Code of Practice for Statistics (the Code) to guide government and public authorities on how to collect, analyse and report on ethnicity statistics. We have been promoting our Standards for Ethnicity Data across the government. We have presented to departments on them and we are getting good feedback. They are being used.
Equality Hub analysts are also working with the Office for Statistics Regulation (OSR) to increase the application of the Standards for Ethnicity Data. This could include making links to the Standards much clearer in the supporting material around the Code. Our Standards might also form part of checks carried out by the OSR to ensure different statistical series comply with the Code. Both initiatives could help reinforce the importance of complying with our Standards.
The final thing we will consider is whether to produce similar standards for other equality groups, such as statistics about disabled people.
Methods and Quality Reports
Our ‘Methods and Quality Report’ (MQR) series looks at different aspects of the quality of ethnicity data and are aimed at more technical users. The MQRs are an important way that we alert users to different data quality issues, and provide guidance based on real world ethnicity data.
The MQRs cover:
- how to interpret and use data
- what we are doing to improve ethnicity data
- data quality issues for specific ethnic groups
The MQRs summarise issues and provide recommendations for users. They are intended as considered and definitive, and are often collaborative pieces. One MQR described disparities in a particular approach to policing – “stop and search” – between black and white people in England and Wales. If geographical differences are not taken into account, these disparities can be misleading. There is also an MQR on why detailed, granular ethnicity data is important.
Blog posts
Our blog posts are another crucial way we communicate important data quality issues. These blogs have recommended the use of harmonised ethnicity classifications through a dedicated series of posts:
- Making ethnicity data more coherent and comparable
- Asking people about their ethnicity
- Harmonising decentralised data collections
Our most popular blog posts were one which outlined the difficulties of comparing international ethnicity data, and another that described the strengths and limitations of attributing ethnicity using a person’s name.
The posts are dynamic and demonstrate progress in data quality improvements. They can provide a sense of narrative, like our posts on why data harmonisation is important.
Wrapping up
Our approach to talking about data quality has helped us address our three main issues:
- Quantity of data
- Data quality and interpretation
- Communicating data quality
Firstly, they address the importance of users understanding quality limitations for different datasets. This is fundamental to interpreting the statistics.
Second, we describe the complexity of some of the quality issues. For example, the MQR on stop and search data describes how the skewed geographic distribution of stop and searches can lead to a counterintuitive result.
And finally, our strategic approach, using different ways to communicate data quality issues through the three methods described here – and presenting on those at events like the Q2024 conference. Our methods take into account the sensitivity of the topic. The more sensitive the topic is, the more important it is to use different ways of communicating quality to demonstrate our trustworthiness and gain respect.
If you have any feedback on our Standards for Ethnicity Data, MQRs or blog posts, please contact richard.laux@cabinetoffice.gov.uk
Richard Laux