When Identifying Risks Creates New Ones: Potential Harms of Socioeconomic and Environmental Risk Indicators as Decision-Support Tools

Abstract: With limited resources and increasing threats to humans and the environment, public-facing indicator mapping platforms aiming to capture overlapping social and environmental metrics are proliferating. These platforms aim to capture, quantify, and combine metrics pertaining to hazards, vulnerabilities, risks, and resiliency, with a broad goal of identifying relative ‘hotspots’ where resources, ranging from public outreach to material investment to further research, should be directed. Although it is a laudable goal, these platforms all face a multitude of challenges to their validity and use, both individually and in relation to each other. I focus on one mapping indicator platform, MIT’s ‘System for the Triage of Risks from Environmental and Socioeconomic Stressors’ (STRESS), and articulate challenges throughout the knowledge production pipeline.

I first describe how attempts to develop STRESS to be a more useful tool for Essex County, MA, 
unearthed fundamental questions about the methodological underpinnings of the platform. I then 
explore three of these methodological considerations both conceptually and empirically: dealing 
with uncertainty; alternative data definitions and datasets; and alternative data normalizations. In a case study aiming to identify hotspots with high heat, poverty, and a large elderly population, I apply a novel method of capturing uncertainty in normalized values, compare several definitions of high heat, and compare two normalizations currently used in STRESS with alternative normalizations used in other platforms. The locations identified as ‘hotspots’ are sensitive to all these decisions that are made, despite the output being mechanically objective. Indicators thus pose new risks of confidently misguiding resource allocation. 

Given these findings, I conclude by suggesting ways to introduce friction into the production and use of these risk indicators. Friction integrates human judgment into indicators, leading to less ‘risky’ indicators.