It’s not evidence-based. Also, they cherry-pick the data sets, not out of malice, but just because it’s too much work to simplify the data sets to be fed into the same model. Every income source of air quality, of water quality, of whatever, was using radically different data formats.