Perhaps it is the case that there is a certain group of students who do well in class and sleep well? The article speaks of correlations at first, then immediately jumps into causation.
Probably because it makes extreme sense to treat, provisionally, in practice and within one's set of tentative hypotesis, that correlation as a causation. Well functioning individuals increasingly behave erratically under sleep deprivation. Inductively "billions" of humans presumably got it as an experience, that rest dysfunction "correlated" with imperfect performance, with clear direction about which side may have conditioned the other. The warning is all on the size of the effect, not on the correlative factors.
Secondary factors such as vicious circles of struggle, anxiety etc. aside.
