in

Three questions to ask before using model outputs for decision support

[adace-ad id="91168"]

Models are developed for a specific purpose and by the need to address certain questions about real systems. Models therefore focus on aspects of the real system that are considered important in answering these questions. Consequently, different models exist for the same system. Without knowing its purpose, it is impossible to assess whether a model’s outputs can be used to support decisions affecting the real world.

Model purposes fall into three main categories: demonstration, understanding, and prediction. Given these different purposes, models also reflect different scopes. Models for demonstration are designed to explore ideas, demonstrate the consequences of certain assumptions, and thereby help communicate key concepts and mechanisms. For example, at the onset of the Covid-19 pandemic simple mathematical models were used to demonstrate how lowering the basic reproduction value, R0, would lead to “flattening the curve” of infections over time. This is an important logical prediction that helped to make key decisions, but it does not, and cannot, say anything about how effective interventions like social distancing are in reducing R0.

Models for understanding are aimed at exploring how different components of a system interact to shape observed behavior of real systems. For example, a model can mechanistically represent movement and contact rates of individuals. The model can be run to let R0 emerge and then explore how R0 changes with interventions such as social distancing. Such models are not necessarily numerically precise, but they provide mechanistic understanding that helps to evaluate the consequences of alternative management measures.

Finally, models for prediction focus on numerical precision. They tend to be more detailed and complex and rely heavily on data for calibration. Their ability to make future projections therefore depends on the quality of data used for model calibration. Such models still do not predict the future with precision, as this is impossible11, but they provide important estimates of alternative future scenarios12.

Decision makers can benefit from all three types of models if they use them according to their given purpose. Modelers should therefore state a model’s purpose clearly and upfront. By asking this first screening question, one of the most common misuses of models can be prevented: using them for purposes for which they were not designed13.


Source: Ecology - nature.com

Validating the physics behind the new MIT-designed fusion experiment

Utilizing conductivity of seawater for bioelectric measurement of fish