This is a really big caveat and I wish it had been highlighted more in the piece.

This is a really big caveat and I wish it had been highlighted more in the piece. I had to delve into the Lindqvist et al article myself to see if they controlled for exercise, since that wasn’t mentioned as a control variable early in the piece. Exercise *was* used as a control in the article, it turns out — but the way they measured it was really really simplistic (three categories: “none”, “walk at least once a week”, and “strenuous exercise”).