Again, this finds a correlation, not causation. There could be some external factor which influences both measured metrics.

Canli as quoted clearly does understand the difference, because he states that one predicts the other, which is distinguishing it from causation. But the article as written implies causation without any justification for this additional leap.