More than 10 years ago, California started using a data tool, called the California Community Environmental Health Screening Tool —or CalEnviroScreen — in order to determine which California neighborhoods were most in need of funding and other services related to pollution and environmental justice.
It’s reported to have directed more than $12 billion in funding for a variety of programs, including affordable housing, public transit, renewable energy, and agricultural subsidies, among others.
But earlier this month, a study that began at Stanford University found that the algorithmic tool has some major flaws.
Firstly, it’s possible that the tool is biased against immigrants: the only “health impacts” it measures are babies’ birth weights, cardiovascular disease, and emergency room visits for asthma. The authors of the study say this metric “underrepresents groups who use the emergency room less, or come from countries where asthma is less prevalent, yet still have other respiratory issues.”
Secondly, the model output was highly susceptible to minor changes in the data. This means that communities that missed out on funding may have been eligible under slightly different measurements, calling into question the “objective” nature of the data and the effectiveness of the funding it determines.
And thirdly, the study found that the tool actually puts communities impacted by environmental racism in competition with each other, leveraging further harm upon them.
The research team suggested using multiple models to offset the subjective nature of algorithms, to include race as a factor in the data, and to implement an external advisory board, made up of domain experts, community leaders and others to oversee its results and provide their own expertise.
A spokesperson for the Cal EPA’s environmental health office says the study’s recommendations are being reviewed.
