On 31st March 2022, the Ada Lovelace Institute held an invite-only roundtable about public attitudes to data, focused on three questions:
- What do we currently know about UK public attitudes towards the regulation of data?
- What are the gaps in our understanding of those attitudes?
- How can we work together to address those gaps?
The event will be written up properly by the Ada Lovelace Institute. I was only in one of the breakout rooms so didn’t hear everything that was said, but do want to reflect on the conversation from a Connected by data perspective. It was held under Chatham House rules, so I won’t be able to attribute the points made in the conversation.
There were a number of useful pointers given at the event to reports about public attitudes to data (alongside Ada’s substantial work on public deliberation). Here are the ones I captured:
- Which?: Consumer data, targeted advertising and user choice is a revealed preference survey focused on Facebook and Google, leading to an estimate of the value of the reduction in harm that would result from giving consumers more control
- Which?: Are you still following me? Consumer attitudes to data collection methods for targeted advertising is a survey of 1,700 Facebook users to explore their understanding of Facebook’s data collection methods and their attitudes towards these methods
- Which?: The Consumer Voice: Automated Decision Marking and Cookie Consents proposed by “Data: a new direction” is qualitative research using a deliberative approach to understand consumer views on some of the proposed changes to UK data protection law outlined in the “Data: a new direction” consultation
- ADR UK: Trust, Security and Public Interest: Striking the Balance is a review of previous literature on public attitudes towards the sharing and linking of administrative data for research
- CDEI: Public attitudes to data and AI: Tracker survey is a survey of public attitudes that includes a conjoint experiment and segmentation analysis to better understand how context affects public attitudes
- Living with data: Public understanding and perceptions of data practices: a review of existing research is a review (usefully also summarised here) of original empirical research into public perceptions of, attitudes towards and feelings about data practices
- Alan Turing Institute: Camden publishes Data Charter promising safer and ethical use of data is a press release about a project in which Camden carried out citizen engagement to help craft their Data Charter
- University of East Anglia: Sustainable participation? Mapping out and reflecting on the field of public dialogue on science and technology maps the environment for public engagement with science and technology more widely than data, across the UK
- Nesta: Outcomes from Data Dialogues used participatory futures methods to explore possible ways health and social care data could be used in future to create greater benefits for everyone
There were a number of themes in the conversation that I found particularly interesting or important.
The importance of context
Perhaps the biggest underlying theme when discussing this research was that context really matters. There isn’t just one answer to “how the public feels about data”, partly because every use has different features that make it more or less acceptable, and partly because there isn’t just one public. We do all have different exposure to risk and risk appetites around data. The recent CDEI research is interesting in that it used a conjoint analysis to try to dig into the important dimensions of the context – looking specifically at the actor involved; the type and identifiability of the data; the purpose of data sharing; and the level of regulation applied – to understand which were most important.
To me, the importance of context means that there is a limit to the utility to these kinds of large scale public attitudes survey, especially when it comes to using them to actually determining what you, as an organisation, should do with data. You need to ask your public about how you’re planning to use data, in your context. And that means we need to develop techniques for engaging communities around more specific data questions.
The gap between stated and revealed preferences
We talked about the so-called “privacy paradox” whereby there’s a gap between what people say they’re concerned about and their actions. For example, given the massive dissatisfaction with Facebook and its use of data, why do people still use it? If we’re so bothered by being tracked across the internet, why don’t we use the controls we have to turn off third party cookies in our browsers? If we want to know more and have more control over our data, why don’t we read privacy notices, or change away from the default settings in our privacy panels?
We discussed the degree to which this is digital resignation, and in particular how our choices are in fact constrained, both by the market (if you were to leave Facebook, what’s the alternative platform that doesn’t have the same privacy problems, but where you can continue to connect with all your friends and family?) and by the choice architectures we’re presented with that nudge us towards accepting the status quo.
The way I am coming to think about this is that we need more collective control over both the market (through democratically determined regulation, for example), and our choice architectures (through participatory governance of how those choices are presented) in order to have meaningful choices as individuals.
The gap between initial and eventual attitudes
Several of the people attending the roundtable reported the experience of working with members of the public in these deliberation exercises and finding that the attitudes they have coming in to the deliberation are very different from the ones they have going out of it. They also reported that people will often suggest that they need things (such as rights or controls) that they already have, suggesting a low level of basic understanding.
People talked about therefore needing to:
- take time over deliberation exercises
- provide a range of different kinds of material that come at the question from different viewpoints and directions
- ground the discussion with practical examples
- give space to ask questions and reflect
- use comparison questions to help work through pros and cons
- be on hand to answer questions
- enable participants to give detailed and specific narrative feedback
- challenge with the consequences of the remedies participants suggest (eg showing them what receiving all the information they say they need would really mean)
There are a couple of things I’m interested in here. First, when should organisations use participatory models such as citizen juries that engage a set of people who are new to the questions; and when should they use models that engage representatives over a longer period of time, such as through boards? Both models have advantages and disadvantages and I have a feeling they might be appropriate for different kinds of questions – fresh stakeholders for the big questions, and established ones for the details, perhaps.
Second, the gap between attitudes going in and attitudes coming out of deliberative exercises concerns me because it means that members of the public who aren’t involved in the exercises themselves may be distrustful of their outcomes. In other words, the external legitimacy of the process – that perceived by those who aren’t involved in it – would be low. I’m sure that this isn’t unique to deliberative exercises about data, and the participation literature will likely have examples of how to socialise the results of these exercises to avoid that risk.
The gap between measurement and action
Most of the roundtable was focused on the measurement of public attitudes using different methodologies. But we know that effective consultation requires more than just listening to what people say; it requires responding to it.
It feels to me that effective participatory data governance could be a practice that builds and maintains trust – actually improving public attitudes over time – as well as simply measuring it, but, conversely, that running lots of exercises whose results are ignored will undermine that trust, and cause even more disengagement.
This is a particular risk when the organisations that are carrying out the engagement and exercises are not organisations that can effect change; or they are but they ignore or suppress the results. Initiatives like the generation of the Camden Data Charter, on the other hand, should be more effective – if the charter is demonstrably followed by the council, of course.
Again, I doubt this is a pattern that is unique to data, and I suspect the participation literature will provide guidance on how to ensure consultation exercises are, and are seen to be, meaningful.
The gap between reality and fantasy
We discussed the role of imagination in how people feel about data practices. One of the findings in the work done by Living with data was that even when the way data systems operated were really clearly explained to people, they didn’t believe they were being given the whole story, and imagined other ways in which data might be being used, now and in the future.
There was some debate about whether being this imaginative was a good or bad thing. After all, one of the things we know about data is that things change all the time – new technologies come along, new algorithms are developed, organisations change hands or change their policies, emergency measures get put in place and so on. So exercising imagination about what might happen is a useful technique for guarding against these futures. Or you can see it as paranoia.
One interesting point here was that some participatory methods can actually use the imagination of participants to create utopian and dystopian futures, which have the advantage of not being shackled to the here and now and therefore can be more radical than typical deliberative approaches. Nesta’s work on participatory futures was highlighted here; the ODI also used a design fiction approach to imagine the future of data portability.
Participatory research into participatory data governance
A final thought from the roundtable was about understanding the field of public attitudes and engagement research more broadly. This spoke a bit to the work we’re planning at Connected by data to explore different participatory methods as they’ve been used in practice, particularly to understand the breadth and effectiveness of different methods being used.
I also suggested that if participatory approaches are being used in data governance – in particular, in the model I suggested above, to shape the choices we are given as individuals – then we should probably query the public acceptability of that as a governance model. This again speaks to how to establish the legitimacy of the kind of collective governance models we’re advocating for.
All in all it was a fantastic discussion and great to see so many organisations grappling with the question of how to bring the perspective of people and communities into decision making about data.