This is the data used in the paper: "Are Anti-Feminist Communities Gateways to the Far Right? Evidence from Reddit and YouTube". It consists of two types of files: Channel/subreddit files Files starting with the prefixes subreddits_ or channels_ are JSON files of the form: { "RedPillWives": [ {"user_id": "rebelarch86", "timestamp": 1546300848000, "category": "TRP"}, {"user_id": "Throwitaway13245", "timestamp": 1546372113000, "category": "TRP"}, ... ] ... } For each channel or subreddit, they contain all users who commented in that subreddit with timestamps. User files Files starting with the prefixes users_ are JSON files of the form: { "/channel/UCsr0bfKhVp_IooMDCmGvDIg": [ {"timestamp": 1532180835686, "channel_id": "UCGDaOZg2INC0Qg2Z203F1dA", "category": "right", "id": "UgxJIPjEJVdZWEYJFIl4AaABAg"}, ... ], ... } For each user, they contain all subreddits (for Reddit-related files) and channels (for YouTube-related files) that users commented on. Gaming vs Random Prefixes _gaming and _random indicate, in Reddit-related files, what counterpart is in the file. For the former, the counterparts are gaming channels (17 to be more precise), while for the latter counterparts are 0.05% of Pushshift's sample. There is redundancy here: each file contains everything + data related to a different control. Other things Code is made available here: https://github.com/epfl-dlab/manosphere_to_altright An online appendix with the list of subreddits is made available here. A list of youtube channels obtained from here is provided in the file sources_final_trimmed.csv. Edit: Upon request on the channel owner’s behalf, we have removed the channel ‘theglassblindspot’, which was incorrectly labeled, from the data. Since the channel is small, removing it has no noticeable impact on the results presented in this paper.