The thing about the article is it's not really offering any air tight explanations. It looks at a significant amount of data and asks what's correlated and what

The thing about the article is it's not really offering any air tight explanations. It looks at a significant amount of data and asks what's correlated and what isn't - many common explanations simply don't fit the data at all. So it comes down to contaminants and psychological effects. But it is not making any definitive pronouncements about which contaminants or how.