Improving participation in environmental decision-making using Bayesian Belief Networks

Abstract

Participation is increasingly recommended to improve democratic processes; however, in practice it is often perceived as an empty and frustrating process (Arnstein, 1969, p. 216). In participatory environmental decision making (PEDM) extractive methods of data collection focus on accuracy of data rather than democratic values and prioritise expert knowledge over public views. As a result, participants become disillusioned and negative feelings develop into active opposition (Wolsink, 2007), causing delays, spiralling costs, and conflict (Haggett, 2008). Previous studies claim that features of advanced statistical modelling, such as Bayesian Belief Networks (BBNs), could aid participation. Using an exploratory case study, a proposed tidal energy scheme in the UK’s Solway Firth, Solway Energy Gateway (SEG), this study investigates how features of BBNs could be harnessed to improve PEDM. A Participatory Action Research (PAR) inspired framework of phases of action and reflection was used to collect and analyse qualitative data from a diverse sample of stakeholders. BBNs were coproduced and co-analysed with the participants over five cycles of engagement. The capacity for BBNs to improve PEDM was evaluated against criteria derived from theoretical ideals of deliberative democracy, procedural environmental justice, and science and technology studies. BBNs were found to partly contribute to the fulfilment of the criteria. Working through the logical structure of the BBNs helped the participants broaden their thoughts and generate knowledge. However, the complexity of BBNs caused anxiety and consultation fatigue. Knowledge was lost as BBNs were built, reducing the ability to communicate knowledge and facilitate learning between participants. BBNs highlighted opposing views, emphasising existing hostilities. There is capacity for BBNs to aid in the scoping phase of PEDM. Further research is needed to explore this proposition and how BBNs could be used in combination with other modelling tools to implement a more comprehensive solution to the challenges associated with PEDM.