I can't agree with this. I don't think scraping specific forums for a small dictionary list of words can be an accurate marker of depression, especially when those forums overlap with your control ones; asthma, cancer, and ptsd sufferers would also suffer depression and anxiety as well, and it's hard to say those would be linked to word usage. It feels more like they had an idea and constructed an experiment to bolster it, considering they have no way to verify what the actual rates of depression and anxiety are apart from posting in that forum, nor any way to take into account lurkers and non-posting participants, who make the majority of a forum's inhabitants.