The article says 1/3 are "at risk" of depression. It looks like the statistic is based on a web survey (https://pdfs.semanticscholar.org/9ac3/478bc1263be53f5150a54b...), so I suppose they couldn't do an actual diagnosis.

Ive theorized the people getting a PhD are not your average american. Most college grads finish school and go into industry.

My peers who went onto get PhDs without going to industry were... different. Not necessarily super smart, but not the best socially or in leadership positions.

I can think of 3 people that were getting their PhDs in engineering, but I thought they were average students at best. One was an average to below average working with me on labs. Another was just a goofy weirdo, I couldnt be friends with him because he'd shoo me away. Another was somewhat a brainiac, but to a fault, He'd get As, but didnt really have friends from what I could tell.

I imagine lots of this is correlation not causation, and that things are different at big names schools.

My experience was the opposite. The people in my PhD program (at an upper-tier state institution) were far better thinkers in my field (biology) than the people in my major at an upper-tier liberal arts college (who, granted, were largely pre-med).

Your article says 6.7% of Americans have a depression in a given year, not that 6.7% of Americans get depression at all.

That can very well mean 1/3 of Americans get depression over their entire lives, and a PhD program would seem like the time to get it.