xfire? LOL... 1 out of every 200 gamers use that app (probably less than that)
Then, logically, it would follow that if you multiplied the # of players and # of hours played by a factor of 200 for each game, you could get an accurate measure of play time for each of those games? Of course, by that metric, the rankings of the games would remain identical.
I believe this would qualify them as a "sample population."
Unless, of course, Xfire users are biased and have a natural disposition towards disliking Diablo 3. Which is, of course, highly unlikely.
It's like the Nielsen ratings for television. They can't sample every home in America, but they can give a pretty accurate picture of what shows people are watching.