Model uncertainty, the COVID-19 pandemic, and the science-policy interface

Area of Science:

• Epidemiology
• Science and Policy Studies
• Mathematical Modeling

Background:

• The COVID-19 pandemic highlighted difficulties in translating scientific findings into policy.
• Managing and communicating uncertainty in epidemiological models presents significant challenges for policymakers.
• Epidemiological models not only represent but also actively shape the uncertainties they aim to convey.

Purpose of the Study:

• To explore the concept of 'uncertainty domestication' within epidemiological models.
• To analyze how models simplify or structure uncertainties for policy applications.
• To assess the implications of model-based uncertainty domestication for science-for-policy.

Main Methods:

• Analysis of three case studies involving COVID-19 epidemiological models.
• Qualitative examination of how uncertainty is represented and managed within these models.
• Exploration of the relationship between model structure and policy needs.

Main Results:

• Models 'domesticate' uncertainties, making them more manageable but potentially less representative of true complexity.
• The process of uncertainty domestication can lead to a mismatch between model outputs and the practical requirements of policymakers.
• Specific examples from COVID-19 modeling illustrate how this domestication can obscure or oversimplify critical uncertainties.

Conclusions:

• Greater attention is needed on how epidemiological models domesticate uncertainty for policy support.
• The methods used to simplify uncertainty in models may not align with the nuanced demands of policy and planning.
• Understanding the mechanisms of uncertainty domestication is crucial for effective science-based policymaking.