To be fair, I think in PEW's turnout model, they had it something like R+2, which is not realistic, although most liberals on this forum told me a few weeks ago that the turnout models don't matter so long as there are sufficient amounts of likely voters, and this poll does have more than the usual 400-500 sample.
Just as NYT/CBS poll from a few weeks ago having Obama up 8 using D+11 was dumb, I dont think R+2 is realistic, either. I dont care about the likely/non likely. The turnout model is what is important. I know its prolly not going to be R+2 and I sure as hell know its not going to be D+7 like in 08. There just isnt that much of a gap in intensity. It will likely settle around the average of D+3 or maybe D+2 since there now appears to be a slight edge in intensity in favor of the GOP.
That said, I would not be surprised if Romney was ahead by 1 or 2 points, or tied. I think 4 is a little high, but I definitely would not be surprsied to see him very narrowly ahead at this juncture.