Abstract for: Trusted Simulation: An Organizing Framework for Considering Model Quality

In an ideal simulation modeling project, the user of the model would be given all the information they need to properly calibrate their trust in the model. The model wouldn’t be perfect, but the user would understand the strengths and weaknesses of the model and could use that information to determine when and how the model should be used. This understanding would incorporate not only metrics characterizing the model itself, but also context in which it is built and used. This article presents a framework to organize our understanding of model quality around this ultimate goal of calibrating user trust. Incorporated concepts include mode applicability, performance, robustness, transparency, and bias, as well as user involvement in model development, model reputation, and usability. This framework facilitates holistic consideration of system dynamics model quality.