Assessing Individual Influence on Group Decisions in Geological Carbon Capture and Storage Problems

Introduction

Subsurface carbon capture and storage requires CO₂ to be stored indefinitely in the intended geological storage reservoir. However, the uncertainty associated with information about the deep subsurface makes prediction of the long term fate of the injected CO₂ difficult. The geological information on which such predictions are based is inherently uncertain, and requires individuals or groups of experts to make interpretations and judgements about the likely properties and characteristics of the reservoir, caprock and overburden layers of rock. A critical component of site evaluation is the analysis of risk where risk is defined as a combination of the likelihood and the severity of impact of any event (Chadwick et al., 2008; Det Norske Veritas, 2010, Smith et al., 2010, Polson et al., 2012). As part of this process, experts are asked to assess the risk of CO₂ storage in a given site using available data, simulation results, and their experience and expertise. Uncertainty can lead to different experts forming different opinions based on the same information (Baddeley et al., 2004; Bond et al., 2007; Lowe & Lorenzoni, 2007; Polson & Curtis, 2010).

Group interactions between experts can improve the quality of decisions made as it allows knowledge and experience to be shared between experts (e.g. Phillips, 1999). All individual experts are subject to cognitive biases when making judgements in situations of uncertainty (e.g. Kahneman et al. 1982, Anderson, 1998, O’Hagan et al, 2006, Bond et al. 2012, Curtis, 2012), and evidence suggests that group interaction may help to mitigate the effects of individual bias (Sniezek, 1992, Diviacco, 2015, see chapter 1 of this book). However group interaction may lead to the introduction of other, group-related biases (Sniezek, 1992), and it is towards these that we focus attention in this chapter.

In situations where a group consensus must be reached, individual experts will exert different levels of influence on the group. In an ideal world the influence of each expert will reflect his/her relative expertise, and group performance will improve with more effective use of each additional expert (Hackman & Morris, 1975; McGrath, 1984). However other factors such as individual force of personality and confidence may bias the group decision. Also the relative influence of each expert may not be apparent from the meeting itself: while particularly dominant characters may stand out, their actual influence on the group may be lower than their dominance of the conversation would suggest. Furthermore even where no expert appears to dominate the discussions or decisions, individual experts may still have a disproportionate influence on group decisions, and on other experts.

By using a novel structured elicitation process we are able to identify when such group biases occur, for example in the assessment of geological information (Polson & Curtis, 2010). Expert elicitation theory and practice have been investigated and used in numerous studies in the earth and environmental sciences such as nuclear waster disposal (Bonano & Apostolakis, 1991), interpretation of geological data (Curtis & Wood, 2004a; Bond et al., 2007), risk and impacts of climate change (Morgan et al., 2001; Arnell et al., 2005; Lowe & Lorenzoni, 2007), hydrological modelling (Ye et al., 2008), and risk assessment connected with volcanic eruptions (Aspinall, 2006) and earthquakes (Bommer et al., 2005; Runge et al., 2013). Using a well-structured elicitation method it is possible to track the opinions and judgements of individual experts in a group environment where a consensus opinion or view is required. Polson and Curtis (2010) used this to track the influence of group discussions on the opinions of individual experts, and to compare the judgements of individual experts with the group consensus. This revealed that a variety of identifiable biases had influenced views in a real geological evaluation.