Tim was an online participant in the hybrid Data Power 2022 Conference.
Due to last minute changes to the Data Power conference schedule I wasn’t able to make all the sessions I’d planned to attend, but did join a few virtual sessions in listening mode (without taking detailed notes), and connected into one panel from Sheffield which is written up below.
Living with Data panel
Panel with: Hannah Ditchfield, Susan Oman, Helen Kennedy, Jo Bates, Itzelle Medina-Perea and Aidan Peppin
I attended a session focussed on reporting back from the latest three year, Nuffield Foundation funded, cycle of the Living with Data research programme, which has been exploring public ‘knowledge, experiences and perceptions of data practices’.
The project has focussed on talking to ordinary people about the use of data about them, and has been interested in how data practice can be improved. It has sought to get into the detail of how specific data practices look, for specific people, in specific places, with a particular emphasis on reaching out to often marginalised and minoritised groups.
Building on an extensive (and incredibly useful) literature review, the project has carried out empirical work through two rounds of a sampled survey of 2,000 participants, and a process of focus groups and interviews. The work has also built on Jo Bates methodology for mapping ‘data journeys’: visualising the ways in which data is collected and used in order to elicit greater public understanding, and scrutiny of, data practices. Notably, a number of the visual maps the team created, covering data projects from the Department of Work and Pensions (DWP), British Broadcasting Corporation (BBC) and National Health Service (NHS) had to include blurred sections because even with the efforts of a research team trying to map out data flows, it wasn’t always possible to discover and document what was actually happening with data between it’s collection and ultimate use.
To pick up just a couple of interesting notes from the talks:
The majority of qualitative research on public perceptions around data that the study team found were in the academic literature, whilst the majority of quantiative and survey research was found in the grey/policy literature.
The survey run by the Living with Data team sought to combine assessment of understanding of data practices, with assessment of attitudes towards them. They found that higher understanding, in general, correlated with lower trust / more negative attitudes towards data uses (the ‘well informed’ cluster identified in the survey were more critical about data use) - but, in one notable case, lived experience of a data system correlated with higher trust. This was where recipients of Universal Credit were more likely to have trust in Department for Work and Pensions data practices, than those who were not directly engaged with DWP data systems on a day-to-day basis.
The qualitative evidence suggested that people from one disadvantaged or minoritized group were often more more concerned about how others from a different disadvantaged/minority group might be affected by data practices. In other words, people engaged in an imaginative leap to consider how data systems might affect others when assessing them. This finding, unpacked by the team in the idea of ‘data imaginaries’, where individuals may draw upon historic stories of data misuse or breaches, or other life experiences, to construct a hypothetical story of future or potential harms of data use, are particularly interesting for indicating (I would argue) a clear capacity and desire for data governance to be treated as a collective, rather than individual, challenge.
One of the outputs of the project, described in more depth in a forthcoming paper from Jo Bates and colleagues, has been a set of ‘Critical Transparency Principles’, seeking to highlight that, if transparency is to be a meaningful part of data governance, it needs to pass a number of tests. Namely, the argue that there are forms of transparency that might enhance collective agency, and to do this transparency must:
- Reduce information asymmetries
- Foster open discussion about what aspects of data systems to make transparent
- Recognise potential and evolving societal impacts of a data system
- Avoid obfuscation when communicating about data systems
- Acknowledge and foster understanding of uncertainty in data systems
- Transparency practice can take place at various data system design / implementation stages
- Recognise resources needed and commit to ensuring they are available
For me, this reads well as an argument for an approach to transparency based both in documentation and dialogue. It’s only by talking with those most affected by a data system that it becomes possible to identify what the most salient information to disclose about its operation may be, and to frame that in a way that connects with the concerns of those affected citizens. There is a question left open here of who takes on the responsibility for delivering this ambitious form of transparency, as it’s not clear it will always be in the interests of powerful data institutions to allow the kinds of more active scrutiny at play here, but this perhaps just highlights the vital need for new models of data governance.
Aidan Peppin, of the Ada Lovelace Institute, was the invited respondent to the panel, and shared a number of observations on the growing demand for dialogue, deliberation and co-design of data practice, and the need to embed more participatory methods into policy making. Aidan also put forward a three-part case for promoting greater public engagement, arguing that it brings greater legitimacy, accountability and evidence into data and technology policy. Sharing experience of translating research into policy advice, Aidan pointed to the vital role that research evidence plays: being able to back up headline points with robust studies. This highlights a key issue for Connected by Data: to be able to move towards a place where we can set out clear evidence on how participatory data governance works in practice to back up our policy asks.
In the final discussion section, I asked a question on whether the research had surfaced any insights into the kinds of ‘collective agency’ that people might be imaginging over data. Ultimately, the panel responded that collective agency over data was more a ‘utopian ideal’ that the research team had projected into some of what they had heard from study participants, rather than something emerging clearly from the survey or focus groups. In most cases, they explained, agency was either cast by participants as something in the data (i.e. data appears to be granted agency distinct from the systems it is aprt of), or agency was explored in terms of individual resistance (e.g. choosing not to use biometric registers at a sports club). This points to a gap between the ‘micro-agencies’ people may feel able to engage with around data right now, and the extent to which people are able, in the current landscape, to successfully imaging having collective agency over data.