Evacuations are the preferred response to human- or natural-caused disasters. The process often involves people deciding when and how to evacuate based on messages from local authorities. However, diverse opinions of the affected people may influence their decision to evacuate or to stay and see how the situation unfolds. This project applies an opinion dynamics concept to model the opinion and decision-making of people threatened by wildfire. To demonstrate how individual opinions evolve with time, the model applies an agent-based approach that includes the interaction between an agency sending an evacuation message and the affected population. There are three sources of information concerning the mathematical model of an opinion: the global broadcasting message, interaction of the agent with its social media network, and observations of neighbors’ actions. The opinion value of each agent leads to a decision to evacuate if it overcomes a resistance threshold. By combining sources of information, the results show that when global broadcasting is the only information available to agents, a decision to evacuate is unanimously reached after a short period. However, when social media interactions are included, there is a delay in reaching a unanimous agreement to evacuate. Furthermore, when social media interactions are replaced by observing the actions of neighbors, there is no agreement to evacuate among the agents, and most of them decide to stay and see how the situation progresses. This research project provides opportunities for planning and management of traffic and routes when an evacuation is expected but the number of people participating is unknown. The results provide valuable insights that could be applied as part of disaster-planning and other potentially life-saving measures.

Publication Date

Summer 6-2024

Publication Type


Digital Object Identifier


MTI Project



Evacuation model, Agent-based model, Opinion dynamics, Modeling infrastructure analysis