More art than science: Modelling energy futures with a Citizen’s Panel

Alice Garvey and Andy Yuille

The EDRC Citizens’ Panel was established to explore the mandate for driving substantial emissions reductions through social change. It is an innovative attempt to bring together citizen deliberation with technical modelling, and to generate scenarios that reflect the views and values of a group who are broadly representative of the UK public. The first themed block of the panel addressed food emissions and what a more sustainable UK food system would look like. You can read more about the findings in a blog and policy brief. Now the block has concluded, we’ve been reflecting on the methodological challenge of trying to integrate modelling with a deliberative process.

The block started with two online information sessions, where the panel heard from experts and commentators, asked questions, and reflected on what they’d heard. This was followed by an in-person day of discussions, where the panel developed visions for a more sustainable national diet, and discussed how they could be achieved. The qualitative findings, and new modelling informed by them, were then fed back to the panel in a further online session after the summer. In what follows we reflect on the experience of trying to get ‘modellable’ data from a deliberative process, and the lessons we’ve learnt along the way.

Perhaps the first challenge was in communicating to the panel what a model is and does. In introductory sessions, the panel heard from a representative for the Department for Energy Security and Net Zero who used the analogy of a model as a ‘recipe’, with several ingredients and a clear output (the ‘cake’). Amongst the questions the panel posed were how can models be made more accessible to the public? Is there a risk that models are biased towards academic ‘elites’ and their worldview? In this way, the panel members pre-empted the very need for the panel in the first place. In the food block we made a clear attempt to document what the model does and doesn’t cover, its underlying assumptions and potential sources of uncertainty.

The central challenge was the tension between the kinds of outputs deliberative panels are good at producing – nuanced and qualitative – and the types of inputs that are most helpful for modelling – bounded and numeric. The problem is how to get information from the panel that is detailed and clear enough for modelling purposes, without asking them to put a number to something fairly abstract or comment on something a food systems specialist would struggle to answer. For instance, a question like ‘what percentage of people in the UK should be vegetarian by 2050’? is not likely to be easily, accurately or meaningfully answered. Integrating public deliberation with modelling involves trying to find justifiable routes towards quantifying qualitative data.

Following established deliberative practice, panel members worked in small groups to develop their own bottom-up visions for change in the national diet, barriers to achieving those visions, and policy and practical measures to support them. To get towards more ‘modellable’ data we created a flashcard activity, with cards representing different levels of ambition for aspects of food system change (for instance ‘where our food comes from’ and ‘the role for alternative proteins’). Panel members were asked to build a ‘scenario’ by choosing a series of cards that aligned mostly closely with their visions. To model the ‘panel-inspired’ scenario we translated this flashcard activity and the more discursive data from the panel into an indicative level of ambition and assumptions around the extent and pace of dietary change. We then compared this to the original scenarios from the Positive Low Energy Futures food model.

Designing activities to carry out modelling from a deliberative process involves a trade-off between ‘relatability’ versus ‘modellability’. That is, on the one hand the topics being discussed need to draw on tangible connections to lived experience. On the other, there is a need to draw out clear quantifiable patterns from the deliberations, which can be scaled up to the population level for use in modelling.

A related tension is the extent to which participants are either constrained in their discussions (for example by having to choose between two options) or can more freely develop their own ideas. Broadly speaking, more constraint meant more modellable outputs, while less constraint meant more deliberative insights. The kind of bridging activity described above was useful in trying to move towards resolving these tensions.

A range of experts – both internal and external – were invited to address the panel, spanning the academic, private and third sectors. A challenge was in how to provide participants with sufficient information to enable an informed debate, whilst avoiding overwhelming them with data. Similarly, ensuring evidence was presented in a form that everyone could rapidly take in quickly emerged as a priority. A surprising development was the level of scepticism that arose even when discussing official statistics, and a suspicion that we weren’t presenting a ‘balanced’ picture.

After outlining the role that livestock play in producing methane emissions, some panel members lamented the ‘cow-bashing’ perceived to be taking place. Some members were critical that one presenter had given global average emissions figures for beef production, rather than UK figures. Similarly, many panel members seemed more willing to accept and believe the perspective of a dairy farmer who spoke to the panel, and this suggested the importance of presenting ‘lived experience’ versus evidence delivered by academics. The literature clearly draws attention to how ‘trusted messengers’ are more likely to be believed or listened to when it comes to messaging around meat and dairy. A distinction is also made in deliberative practice between input from ‘experts’ and from ‘commentators’; however, the boundary between the two may not always be well-defined.

A further surprise was the degree that some panel members wanted to hear more from ‘influencers’, rather than experts or officials. Many panel members lacked trust in government, for instance when it came to communicating advice about sustainable diets. To some degree this explained the difficulties we had in presenting government-produced data, given this lack of trust in government as a messenger. In this way, the panel highlighted tensions between public and academic perceptions of what constitutes reliable, meaningful and authentic information, both of which have valid and flawed aspects.

Whilst challenging, the process has to some extent opened up the ‘black box’ of modelling to a public audience and is making tentative steps to providing more socially acceptable, equitable energy demand reduction scenarios. Given the heatedness of popular debate around veganism for instance, it’s fair to say the topic of food was approached with some trepidation. However, it is an engaging topic for most people, which everyone necessarily has a stake in. Panel members engaged enthusiastically in discussions, even while some were critical of some of the information they received. Such enthusiasm should not be taken for granted in a process which requires a long-term commitment of free time from panel members.

A further block on materials and products has taken place, and future blocks will cover housing and transport, before combining sector-specific modelling into whole-economy systems modelling. We are iterating our processes to best engage panel members with a range of highly complex – and often controversial – topics, to develop modelling that is more sensitive to social intelligence. We anticipate many more challenges and lessons to be learned!