Reconstructing decision-making dynamics during public health crises by applying data science to public records

Public health policies can dramatically shape government responses to emerging public health crises. In cases where a response is inadequate, it's natural to seek improvements to these policies to achieve better results in future crises. However, while policies and outcomes are usually visible, the complex dynamics that link them are seldom obvious. Obscure social network structures, power and information asymmetries, and political agendas all influence the translation of policy to action or inaction. Intergovernmental communications are often the only primary sources for researchers investigating these hidden but crucial factors. Public records laws in many countries make such documents available upon request, but their structure, format, and scale are rarely accessible in practice. Fortunately, recent advances in free and open-source data science tools are making this problem tractable. In this visual brief, we demonstrate using such tools to mine a large image dataset and reconstruct decision-making during the Flint Water Crisis in Michigan, USA.