From 82games:
Stat ON Court OFF Court Net
Offense: Pts per 100 Poss. 93.2 103.6 -10.3
Defense: Pts per 100 Poss. 89.6 105.8 -16.2
Net Points per 100 Possessions +3.7 -2.2 +5.9
A big difference both defensively (for the positive) and offensive (for the negative). Net Net though, he is still making a positive contribution but statically it is small.
What I see on the court is a good player but not so good he should make this difference. His effort is contagious but it is not like they weren't playing good defense before.
Yes, while Bradley has been something of a defensive spark our defense was clearly showing improvement before he came back. In late November the Celts were 23rd in defense, in late December we were 11th. While the team was playing good defense in fits and starts we went through 3-4 game stretches where our defense was as good or better than it's been recently.
In terms of his on/off numbers, that includes all of the games when we were playing bad defense (and better offense). For instance, our defense with Bradley on the bench the last 7 games is (ballpark) 97, so his defense would be closer to -8 than -16.
Overall, according to nba.com, Bradley is a +18 over the last 7 games and he's playing almost 24 minutes a game. Over those 7 games we're a combined +54, so in the 24 minutes a game Avery plays we're a combined +18, in the 24 minutes he's on the bench we're a combined +16. So if you want to attribute our recent play to Avery and not to the team (as a whole) playing better you need to claim that the "Bradley effect" extends to how the team plays when he's on the bench.
And, for the record, I'm a AB supporter, I've always claimed that the Celts with Bradley would have beaten the Heat with Bosh, and I think he's going to be a pretty good offensive player.