How much of "the Reddit" did you actually use to calculate these stats? Is this based off a one-time snapshot from the API (which limits to 1000 posts, per type of query) or is it a longitudinal crawl (and how data was produced in that sample)?