The causality could also go the other way. The most visible users will be those who post a lot. And those who are depressed because their life sucks and they

The causality could also go the other way. The most visible users will be those who post a lot. And those who are depressed because their life sucks and they don't have much valuable stuff to do, or those who aren't doing much valuable stuff because their depression makes it too hard, are likely to have a lot of time during which posting on the site seems like the best available activity. It seems really difficult to separate causality here. Which is something that disappoints me about the article. The string "caus" does not occur on the webpage; three of the four bullet points say "associated", and the only bullet point that uses a causative-sounding phrase ("were protective of") sounds like it may be used to mean another statistical association rather than an established causation. Clicking through to the study, it seems to be 100% based on association of results that they got via surveys.