Names of communities, users and content removed from dataset and not analysed. Unique user IDs MD5 hashed. Data collected via web scraping - conducted only on public data, and done slowly enough so as to not interfere with the service being scraped.