Every time we go to the doctors, take an exam at school or pay council tax, data is collected about us. This information is mostly used by the organisation who collects it to provide a service, for example to maintain an individual’s medical record to treat them appropriately. As more and more data is collected, researchers have been able to harness this new and powerful source of information: routinely-collected, or ‘administrative’ data. You can find out more about administrative data through watching this short film. These datasets can tell us about trends in the population. Through securely linking de-identified health and non-health data we are also able to answer questions about rare diseases, and groups of people who are not usually captured in traditional research studies. 

Alongside this has been a movement to increase the participation of communities being represented in the production of such research. Coined by disability rights movements, the phrase ‘nothing about us, without us’, has increasingly been evoked in relation to research, with many funders of data-driven research now insisting that ​​​​public involvement and engagement (PIE) is carried out prior to submission of research proposals. In qualitative work and interview-based studies, this often involves discussing the questions to be asked of participants, and aspects such as recruitment and dissemination. ​​But what does this look like when it comes to administrative data studies, where the researchers will never know the people behind the data, and analysis techniques are often incredibly complex? 

How can we, as researchers and engagement professionals, go about involving people from under-represented groups in administrative data research​?​

Involving people with lived experience makes administrative data more useful…

Our experience tells us that the way we work with people is best guided by the participants themselves. However, ​for people who are new to this work h​aving a range of options can help to both guide participants and assist researchers when designing and costing PIE activities. We have therefore pooled some of our more successful ways of working together below:

Project design

​​Insights gleaned at the design stage may influence what variables are used in analysis​. F​or example​,​​​ care experienced young people in Northern Ireland changed the focus of a research project to looking at positive, rather than negative, outcomes associated with care experience. They were tired of research and reports that focused on negative experiences and outcomes, and wanted a more holistic representation of the lives of people with care experience. They worked with the researchers to identify new priorities and rewrite the title to ‘Positive Trajectories for Care Leavers’.  ​In a different project, discussions were had with ​people who use drugs about which services they do, and importantly don’t, access​. The insights​ made researchers think more about missing data and how they handle that. ​     ​​

​​T​alking to people about their experiences and perspectives can be incredibly helpful and can shape the way you think about your project, as well as how you pitch it to funders. How you go about this depends on a range of factors including how well the group knows each other, what they usually do, and sometimes, how sensitive or tricky the subject matter is. Projects have used techniques including written outlines of proposed project plans and verbal gathered feedback for those used to talking about their experiences verbally, whereas others have used visual techniques for groups less used to that process.  Visual techniques include using traffic light systems to categorise potential research questions into three colours: green (let’s do this), amber (maybe) and red (let’s not bother), and creative methods such as collage and drawing to explore how people felt about a sensitive subject.

Analysis and interpretation

This can be the most difficult part for administrative data projects, due to the often complicated and technical nature of statistical analysis. As such, many researchers are put off involving members of the public at this stage.  

Even if ​the public​ cannot directly analyse data, there are still ways for them to be involved in scrutinising the data and intermediate results, particularly as it relates to their own lived experience. For example, when working with care experienced young people in Northern Ireland, researchers would bring initial results to the young people as soon as they were able to and facilitated discussions with them to better understand the stories behind the data. This sometimes led to further questions being asked of the data based on the young people’s analysis of those emerging findings, without them having to interact with the data directly. 

Discussing emerging findings can offer insights which only people with ​lived experience ​can provide. This is particularly the case where the results seem odd, or perhaps counter-intuitive. An example that particularly stands out is around an emerging finding that smaller schools in a Scottish city had declining levels of mental health among its children, the opposite of what the team anticipated​. D​iscussions with staff revealed that these schools were likely to permanently close, resulting in selective attrition through some families removing their children prematurely, and a perceived decline in staff morale. By enabling these discussions to happen as the research progresses, analyses can dig deeper into particular aspects, local contexts and meaning behind the data.

Dissemination 

No-one outside of academia wants to read a journal paper, nor are they typically written in a way that is easy for anyone beyond academia to understand. But this does not mean that people are not interested in what the research has to say, especially when it directly affects their lives. Co-producing dissemination materials means that your research can get out to the people that need to hear about it most. Again, this should be tailored to your participants, but other projects have used:

• Animated film techniques, which involved participants designing storyboards, creating and filming the animations, and recording voiceovers: [My data: an animated film, co-produced with people who use drugs — Discovery - the University of Dundee Research Portal ] 
• Story-telling, or creating blogs which relate findings from administrative data research with experiences of people’s own mental health journeys [Mental Health – Imagining Beyond – A collaborative project between The Community Wellbeing Collective, Westerhailes and Mental Health Data Scientists at the University of Edinburgh. Funded by Research Data Scotland.]
• Re-writing of journal articles in more accessible ways: The mental health of all children in contact with social services: Young person's version of a research paper

But this all sounds quite challenging…

The biggest hurdle for many people in starting this type of work is taking that first step and working out how to access the right people. ​W​hen working with under-represented groups, we have found some of the most successful partnerships happen when working with a voluntary, community, or social enterprise organisation, such as a charity. Be brave and reach out to such organisations – we have found that they are generally pleased to work with researchers where the intentions are good. Working with a partner organisation who already holds relationships with the people you want to work with has several advantages: 

• staff often have a lot of knowledge about the subject and community, which can help with discussing approaches and materials ahead of talking to people;
• they may have premises or online forums where people already meet, which you may be able to use, making the experience more comfortable for the people you are involving;
• voluntary, community, or social enterprise partners can often distribute incentives more easily than university finance systems; and they can play a safeguarding role, where you are working with very vulnerable groups. 

However, there is a lot of trust involved in organisations letting researchers build relationships with their members, and researchers need to tread carefully. Researchers will have several meetings with staff members of the organisation before meeting people attending their services and run all parts of the proposed project by them. This has changed the way we worked with some groups: for example, when working with Restoration Fife, a recovery organisation for people who use drugs and alcohol, the University of Dundee team were advised that incentivising individuals could put them at risk of victimisation through receiving large amounts of money, so we worked with staff to ensure that incentives were broken down, with some money given to the whole group, to ensure people were renumerated for their time in an appropriate way. While there is guidance available to support public involvement and engagement, this situation was not considered in ​the ​NIHR Incentive guidance, and the team therefore needed insight from Restoration Fife to get it right. 

Another challenge is ensuring that your participants have the right sort of skills to work with the complexities of administrative data projects, especially if they might be involved in the analysis and interpretation elements of the project. This might need more substantial input from the research team in terms of time, money and training. One way ​to do ​this more intensive work is to bring people with lived experience into the project as staff members. The University of Dundee is currently trialling the recruitment of an ‘expert by experience’, who is being employed two days per week (staying under the 16-hour limit which may have implications for benefit payments) for the duration of the project, to assist with all elements of the project design, interpretation, public involvement and engagement and dissemination. Alternatively, Administrative Data Research Northern Ireland (ADR NI) have put in place a successful internship programme for care experienced young people who have been working with researchers​​. ​​Over an eight-week period in the summer, the intern works eight hours a week with ​​​​researchers and engagement staff, supporting the development of accessible research outputs and guides, engaging communications materials, and designing PIE workshop outlines, as well as receiving mentoring and support from Queen’s University Belfast’s Widening Participation ​​​​Unit. The internship not only helps make data-driven research more accessible and engaging, but also enables the voices of the people represented in the data to be reflected and heard within the research. As the first ADRC NI intern said in a ​​blog reflecting on the end of her experience, “I knew I could help young people understand how data research works, and that young people can do the jobs that the adults do, and that young people can make their mark and have their voice heard.” You can read the blog here..

Money, money, money…

​​But of course, none of this is possible without money. This is a real challenge, particularly for early career researchers, and those working in new fields of research, where they don’t have established relationships. Researchers can look out for specific funds which can be used for PIE activities, and to support building relationships with particular groups. These can be external through research councils and charity organisations, or through internal funding opportunities within institutions. We have previously had success through ‘piggy-backing’ on other large studies, which have PIE groups and kindly let us speak to their participants as a one-off prior to proposal submission, or through charities, which may let you take up part of a meeting, but this method is not sustainable. ​​​

​​I​f PIE is to be valued within data-driven research, it’s critical that the funding landscape around research reflects this. The good news is funders are increasingly recognising that public involvement and engagement is best embedded through a project and requires sufficient costing including buying in expertise such as animators, artists, musicians; travel costs; equipment; incentives; food for participants; printing and design.   

Conclusions

We are privileged to have a huge amount of data available to research: using it in an inclusive way is not only ethical but also leads to more valuable impacts. ​​Involving people from the start of a research project can have immeasurable impact on science and increase trust in research using public sector data. For researchers, this type of work can inspire ideas and be truly fulfilling. Whilst for members of the public, involvement in big data projects can help people to develop new skills and understanding.  Most importantly, though, it can ensure research accurately captures and reports the experiences of the people behind the data.

A PDF VERSION OF THIS ESSAY IS AVAILABLE HERE

---

This work is licensed under CC BY-NC-ND 4.0  

Disclaimer: The views expressed in this essay are not representative of the views of the British Science Association or UK Research and Innovation.