Because you need multiple seasons of defensive data to get the same level of r2 (generally 3, but 2 may be OK for middle infielders b/c of the increase in chances/9), you should basically think of one inning of UZR data having about 1/2 a PA of predictive value. I've written about it here and its been written about consistently on The Book's blog, so I assume b/c you are into UZR you would be interested in that wealth of discussion. (FWIW: If you'd like to read the older discussions of UZR the best discussions were on FanHome's old sabermetric message board way back in the day, but the discussions on BTF in the beginning were interesting, and the discussion on The Book's blog are technical and interesting. MGL's chat at the same time as Dewan's SoSH chat was good, though MGL as always loves to write in snark. The better discussions were also when DRA was introduced and compared/contrasted with UZR, but that must be at least 6 years ago now.)

To put it this way: if you want to predict someone's UZR for 2010, you really need two to three years of data for the equivalent of ONE year of OPS data. Therefore, while Jed's career fielding data (spread over two years), is the equivalent of about 1/3 of a season, that has the predictive value of about 1/6 of a season's worth of PAs. This discussion has been had over at The Book's blog repeatedly.

I use a lot of shorthand as I try to make points, so I deserve it when people snarkily tell me that I should use PAs rather the ABs or otherwise. So, I invite it I suppose. I should do a better job describing my point.

My basic point is this: Jed's UZR data in about 1/3 of as season spread over two years is virtually worthless in predicting his next year's UZR. And, as the author of the fielding bible said recently (check the SoSH board, I'm happy to provide a link if you'd like), you should probably use defensive statistics for no more 60% of your defensive assessment even if you have good data. Therefore, I'd come close to ignoring Jed's data completely. But, I'd certainly, under no circumstances, say that because Jed's sample size is small, it nonetheless is directionally useful. Either the data set is small and therefore worthless, or its of appropriate size and therefore worthy. To put it in terms of math: your error band covers way too much ground (in looking at predicted 2010 UZR's) and would likely be -50 to +50, or -5 to +5 if you regress to league average enough. In all events, he isn't significantly above average or plus according to UZR properly understood. And observationally, I think we agree: he looks about average when you look at the full package. I don't think his 2010 expected defensive contribution should be much better then average until there is more data.

(FWIW, again, I find minor league defensive statistics to be worthless for now until I can see their predictive merit. I find them fun but pointless frivolities. So any data from Jed's minor league career is particularly pointless. We can discuss why if you'd like.)