What does good practice look like and what are the conditions that enable inclusivity to thrive?

One of the questions the Taskforce considered was about how we can learn from experiences here in the UK and more widely in improving our approach to equalities and inclusion going forward. In this section, we highlight examples of promising practices that have been shared with us by participants in the consultations, or that we are aware of through our own experience. We hope they may help to demonstrate some ways in which our recommendations can be put into practice.

Building trust through engagement

An important message we heard from participants in our consultations was the need for a trustworthy system, supporting all groups across society to understand both the benefits and the risks of data sharing.

In the UK, the production of government statistics is underpinned by the Statistics Code of Practice requiring statistics producers to think about the “rights” of those people whose data are being collected, shifting the emphasis from the data to the individuals providing their information. Alongside the Code of Practice, the UK statistics regulator, the Office for Statistics Regulation, has issued guidance on building confidence in the handling and use of data. This urges producers to proactively consider the rights of people providing their data to support the public good. Respecting these rights is crucial to protect and empower citizens within this exchange.

There are numerous good practice tools that statistical producers can draw upon as well. For example, the Open Data Institute: Data Ethics Canvas visually distils the key ethical questions that researchers should consider.

Answering the right questions

To build an inclusive statistical system, roundtable discussion participants highlighted the importance of having strong foundations. Good practice was identified in how analysts in government see themselves, understanding that their primary role is not to produce data and evidence, but to work with others to identify and answer questions that are important to society. This is a subtle but important distinction. From this foundation it is easier to ask ourselves whether people or experiences are missing in the way in which those questions are answered.

Looking beyond the UK, the notion that statistics add most value when they answer society’s questions was echoed in a review of measuring social exclusion by Statistics Canada and the National Institute of Statistics and Geography of Mexico. The researchers highlighted that in developing indicators of social exclusion, it is important to understand and address the main social and policy questions from the start.

Engaging all communities

The extensive community engagement process that goes into delivering the censuses across the UK, is another example of good practice. Participation is encouraged by creating networks of organisations, community leaders and charities that facilitate direct links to relevant groups and populations. Engagement is also undertaken with members of the public directly to ensure a balanced view of concerns and needs. Developing knowledge and understanding about what matters to people has helped researchers to communicate more effectively about the importance of the data they provide. Working with community experts who can advise on the best channels to communicate with local groups and the language that would most resonate is also very important.

The Government Statistical Service has also published a new User Engagement Strategy, to provide practical guidance on user engagement. The strategy emphasises that: we don’t need a few people thinking about inclusivity (or user engagement) perfectly, we just need lots of people trying their best to do it well and working together to make it happen as part of business as usual.

Partnerships

Partnerships work well when they bring diverse people together. This can challenge us to think differently and set higher expectations. In Colombia, the national statistical office has created a multidisciplinary group to help mainstream an intersectional approach. Members include statisticians, economists, a psychologist, an anthropologist, and other advisers. This group has started developing data disaggregation guidelines and piloted questions to measure sexual orientation and gender diversity in surveys.

Inclusivity in routine data collection

When undertaking surveys, the standard sampling frame is households. However, it is equally important to consider non-household populations such as residents of communal establishments (for example, care homes, prisons), Gypsy, Roma and Traveller communities, and homeless people. An example of comparative approaches to these issues comes from Australia. Here effort is made to include non-household residents using a list sample of non-private dwellings such as hotels and motels. Similarly, the US equivalent of the Labour Force Survey (the Current Population Survey) aims to include non-household residents; the stratified sampling frame includes a “group quarter” stratum containing those housing units where residents share common facilities or receive formal care.

Developing new approaches to inclusive data collection

Statistics Canada’s efforts to be inclusive and timely in providing insights into experiences during the pandemic have been highlighted in a public policy forum looking at innovation and leadership during the pandemic. This showcased how Statistics Canada and the Canadian Chamber of Commerce worked together on the design of the  Canadian Survey on Business Conditions. Their approach was to crowdsource data, using email to get people to answer an online survey. This was launched within days of the economic shutdown to collect “real time” data on the pandemic’s impact. Statistics Canada put a special focus on how minority-led businesses were coping, asking business owners to identify the percentage of the business owned by race, ethnicity, sex, whether they are Indigenous and whether they identify as lesbian, gay, bisexual, transgender, queer and/or two-spirited. Patrick Gill of the Canadian Chamber of Commerce said:

Statistics Canada also engaged with disabled people for a crowdsourced survey on how those with disabilities were faring during the pandemic. It worked with organisations including Children First and Vanier Institute of the Family to understand experiences of parenting during the pandemic. Statistics Canada will be developing principles on the appropriate use of crowdsourced data to share promising practice with others.

In the UK, the Government Equalities Office ran the National LGBT survey in 2017 to gather information about the experiences of lesbian, gay, bisexual, and transgender people in the UK. The survey response was unprecedented and over 108,000 people participated. This provided  rich insights into experiences of these groups in the areas of safety, health, education, and employment.

Making the best use of existing data to improve inclusiveness

A good example of the use of administrative data and data linkage for improving inclusiveness is the Scottish Government’s health and homelessness in Scotland project. This linked local authority data on homelessness with NHS data on hospital admissions, outpatient visits, prescriptions, drugs misuse and National Records of Scotland information about deaths to explore the relationship between homelessness and health in Scotland. It included transparency around the risk assessment process which enhanced Scottish Government’s trustworthiness to those involved in sharing and using the data. They published their data privacy impact assessment alongside the main analysis report, including the original application for the data, how it would serve the public good, details of the application’s approval and how others could access the data. This approach is now standard practice for all Scottish Government publications based on linked data.

Another example of using existing data to provide more inclusive insights is ONS’s Coronavirus (COVID-19) related deaths by ethnic group, England and Wales. Ethnicity is not recorded on death certificates and therefore to undertake this analysis, deaths involving COVID-19 were linked to the 2011 Census for England and Wales. This allowed ONS to produce statistics on mortality due to COVID-19 by ethnic group, revealing which ethnic groups were at greater risk of dying from COVID-19, and helping to identify groups that have been disproportionately affected by this disease. Linked data have also been used to explore Deaths involving COVID-19 by religious group in England.

Inclusive dissemination and communication

Trust and transparency can be improved by more effective feedback. Good practice in communication of how data will be used or why suggestions were not adopted, was highlighted as a way to improve participation in, and inclusivity of, research.

This shows the importance of communicating effectively with prospective research participants, by demonstrating how the confidentiality of their data will be maintained and in providing feedback on how their data have been used.

The ZOE COVID-19 symptom tracker app, created by a collaboration of academics, including King’s College, London, is another example where engagement is enhanced through feedback to participants. Users of the ZOE app regularly report on their health and symptoms and whether or not they have tested positive for the virus. Participants are provided with updates and alerts which show how their data is shaping the latest coronavirus evidence. The tracker helps provide real time intelligence on the scale of COVID-19 outbreaks and how it affects different demographics and the information is also shared with the National Health Service.