I looked at about 60,000 distinct users. But you're right about my overall strategy. I chose all of the subreddits with over some number of subscribers (I forget what the number was now.)I ended up with 433 subs. I filtered out the current default subreddits from this visualization.
One thing I was wondering in terms of reddit research - have you looked into this at all - is that they have users check a specific box if they are ok with their voting data being used for research - even if it's already public. My question then is this - is it somehow wrong to use (already-public) data for research? Anyway, I talk about my original aims for the project in some other comments.
Thanks for the link to stattit. My strategy for getting enough threads for my other project was just to keep a slow scraper running for a month and then go back to it - stattit will be incredibly helpful.
> One thing I was wondering in terms of reddit research - have you looked into this at all - is that they have users check a specific box if they are ok with their voting data being used for research - even if it's already public. My question then is this - is it somehow wrong to use (already-public) data for research? Anyway, I talk about my original aims for the project in some other comments.
Based on the dozens (at least) of papers published each year that use twitter data, I'm pretty sure it's kosher to use public posts. You might want to double check with your irb though. Depending on how you present the information, so users might be concerned about their privacy - I wrote a bot that replied to people posting variations of 'your comment history' with a link to the referenced person's redditgraph and several people said they were creeped out by it (a little more here, if your interested: http://www.roadtolarissa.com/redditgraphs-retrospective/).
Depending on what you are looking for the rate limit might slow you down a lot; you might want to contact the site admins:
> tl;dr If you need old data, we'd much rather work out a way to get you
a data dump than to have you scrape.
One thing I was wondering in terms of reddit research - have you looked into this at all - is that they have users check a specific box if they are ok with their voting data being used for research - even if it's already public. My question then is this - is it somehow wrong to use (already-public) data for research? Anyway, I talk about my original aims for the project in some other comments.
Thanks for the link to stattit. My strategy for getting enough threads for my other project was just to keep a slow scraper running for a month and then go back to it - stattit will be incredibly helpful.