Well, it’s comforting to hear about those low percentages (10-20%, as you mention) in your experience. However, empirical results [0] have demonstrated percentages of mental illness among PhD students more in the range of 43-46%. Lest we assign some sort of “well, it’s better here” thinking, those results are from Berkeley.

I remain pedantic about this only because, despite some anectodal experiences, there’s a serious problem in academic training which is having strongly detrimental effects on student development, quality of life, and retention.

[0]: https://www.nature.com/articles/nbt.4089