People play what they own, so if I would have done such an analysis I would have tried to incorporate (if possible) somehow the number of sold games as well, or maybe analyze the libraries of players and see the ratio of newer games to older to try and assess the percentage of "new" games per player on average. if you own 5% new games then it makes sense you play them less than older games since I would speculate that many (most?) players like to switch between games.
But considering the numbers, I'm most definitely one of those playing mostly old games. Sometimes when I buy a new game _for me_ it's already been out a few years and is thus no longer considered new on the market anymore.
Edit: Another thought that came now. I have what I would call "comfort games", it's games I play when I land between other games. Let's say I buy a new game, I try it and like it - I play it, sometimes even finish it, then I go back to my comfort game. Then maybe I try another new game, don't like it and go back to my comfort game while searching for something new. If this comfort game is old, then those hours add up in a year and despite buying maybe three new games, the total hours spent would be in favor of my old, rusty but beloved, comfort games. For me, those are Europa Universalis and the historical Total War games for example. And yet another thought while I'm rambling away, many opl play socially so that ie Counter Strike is still high up is not that surprising if compared to single player games since after you completed a single player game, most ppl never play it again (but of course some do) where-as in an MP game there's always another match to play and it doesn't get repetitive in the same way. So this too will skewer the statistics on games genres vs genres if looked at combined. Statistics is fun, you just gotta think ahead a bit and plan what you want for a result before you collect and analyze your data.
Edit2: Hah, one final thought no one probably cares about but anways: in MP I assume time spent in lobbies count as time as well, so 10000 players that spend 5% of their time in lobbies still add up a lot of hours vs SP games where you have some loading times perhaps and other idling but likely less?