The numbers are normalized for the league in general. The quality of basketball improved as the year wore in (not a surprise, really, it takes time for most teams/players to get in the groove) - and Roy's numbers reflect that.
The deviation isn't that big of a difference anyhow. Variable statistics are a relatively new concept to the NBA, and at times it's hard to break old habits of relying on static statistics.
The league average stats have improved while he's been out. So his stats have gone from above average to below average.
Nice to see Mills so high up there. The guy is really coming along. He may be the second coming of Damon Stoudamire (a player I hated more than any other) but for his role he's a nice fit. As he gets more comfortable I can really see him becoming a 15-16 PER player.
He played 43 minutes or so, shot something like 42%, had a lot of turnovers, genereted few assists, steals or blocks and just wasn't very efficient -- and that's what the metric is measuring.
This is why PER shouldn't be considered alone, of course. His PER for last night was 8.1 - all it means is that he was inefficient, not that his performance was poor or that he didn't help them win the game. When you think about people with low PER's, know that it is not a be-all/end-all for performance, only efficiency. For example, Wesley Matthews is inconsistent (thus his average PER) but is an overall asset and obviously worth the money because when he's on, he's ON FIRE.
