Originally Posted by
SkulFuk
There's a problem with your method - you're missing everyone who's got their account set to not show in that list. For example, right this second there's 7 people online set like that; none of which you have listed.
There's a huge technical problem with my population of interest in general. Right now, the actual population I'd be testing is "Members of the toribash community that were visibly online at undocumented irregular intervals during which I updated names" and is the only group of people I'd be able to infer about with any accuracy, statistically speaking.
My method wouldn't yield results that are valid enough to make it into breaking news, by no means. Like I said, it's good for shits and giggles.