Investigating the Key Parameters of an Agent-Based Model of Mining Community Preferences for Managing Social Risks

"Mining companies have started using quantitative tools, including computer models of community interaction, to gain intelligence on social risks surrounding their projects. Models of changes in community preferences regarding mining projects over time are useful for evaluating changes in project risk due to changes in the social license to operate. For example, agent-based models that use information diffusion models and social networks are useful for studying those changes due to information diffusion. However, such agent-based models are sensitive to many input parameters including the parameters of the diffusion model and the average degree of the social network. This work evaluates the sensitivity of such a model to diffusion and network model parameters (probability of imitation, probability of innovation, and average degree) using the first order and total sensitivity indices. The results show that the model is much more sensitive to the probability of imitation than the other two parameters. Thus, to reduce uncertainty surrounding the model’s predictions of community acceptance of mining, mines need to obtain accurate estimates of the probability of imitation. INTRODUCTION Mining companies have started using quantitative tools, including computer models of community interaction, to gain intelligence on social risks surrounding their projects. Models of changes in community preferences regarding mining projects over time are useful for evaluating changes in project risk due to changes in the social license to operate. For example, agent-based models that use information diffusion models and social networks are useful for studying those changes due to information diffusion. However, such agent-based models are sensitive to many input parameters including the parameters of the diffusion model and the average degree of the social network. This work evaluates the sensitivity of such a model to diffusion and network model parameters (probability of imitation, probability of innovation, and average degree) using the first order and total sensitivity indices. AGENT-BASED MODEL OF COMMUNITY ACCEPTANCE A few researchers have started using agent-based models to study mining systems (Fujiono 2011; Nakagawa et al. 2013; Bahr 2015; Boateng and Awuah-Offei 2017a, b). Nakagawa and co-workers used ABM to evaluate dynamics of a mining project’s stakeholder networks whereas Fujiono used ABM to study innovation diffusion in the mining industry. Our previous work has presented models and a framework for modeling information diffusion in a mining community and its impact on level of acceptance (a measure of SLO) of the mining project (Boateng and Awuah-Offei 2017a, b)."