I don't trust surveys that break down the numbers as far as particular models, but I put some weight on numbers for an entire nameplate simply because the sample size is so large at that point that most of your bias-induced variation is going to work itself out. There will always be some bias in the numbers no matter what, but this bias has been proven to favor the Japanese automakers moreso than the Domestics, so if the point can be made that Domestics fair just as well in surveys even with some moderate bias in there, it's only that much more valid if it was possible to remove it.
Those numbers also pan out over a number of different benchmarking companies with some minor variation in ranking. I simply used JDPower's because I readily found it with Google.
