What a deeply troubling article! a) Typical situation of causation is not correlation. b) Using Dropbox data from 400.000 academic users without consent? They say the data was anonymised, yet contained folder structure, institution, and researchers' seniority. WTF!https://twitter.com/sandeepzgk/status/1020368430745059328 …
How could you possible hope to anonymize data which includes every Dropbox folder name of a participant? There are only so many researchers in each field, with that much data it'd be trivial to reidentify study participants