Abstract for: Perfect Mixing in COVID Models
Most models in System Dynamics, and many in epidemiology more broadly, use aggregate difference of differential equations which means that all stock values will have perfect mixing. While it is true that this is not what happens in the spread of a disease, the important question is whether this matters for the use of models in understanding and the methods used to compare model results with measured data. In this paper perfect mixing models are contrasted with conveyor-based models in which material enters and leaves without mixing. These two types of models are compared against the outcome of an agent-based simulation which is taken as the ground truth. In general, the conveyor base model is more like the ground truth model both in terms of local behavior and the interpretation of the underlying parameters. However, the two model types are closer to one another than they are to the agent-based models. This suggests that straight comparison of measured data to model results for aggregate models should be approached with a great deal of caution.