A D+4 sample is probably close to accurate: shifting 2 percent of the sample form D to I (which is probably a little more accurate) may give Romney another percentage point, but the difference it makes isn’t just less than the margin of error, it’s within rounding error.
Rasmussen does regular party ID surveys to update his models, and his last (about a week ago) was exactly opposite: R - 38%; D - 34%
That’s a huge swing.
Hank
I like a D+4 sample. If Romney’s up in that, then it looks better and better. Of course I’m not a fan of ARG polls. They pale compared to Rass.
Yesterday, there was a thread stating that the party ID had changed to record highs for republicans. It was so high, there were more republicans than democrats. On top of that, if anyone thinks that the likelyness of democrats to show up at the polls being higher than republicans, they need their political heads examined. The enthusiasm for Obama is way down from 2008, and even though Romney wasn't everyone's favorite candidate, he carries more enthusiasm than McCain did.
Add up all the factors, and consider the fact that if you didn't vote for Obama last time, you aren't going to this time, and Obama is toast.
The only polls that count are those state by state with the large electoral votes. Unless we can see those states at least breaking for the GOP at least by 50% or more, this means zip. A national popular poll is useless unless it comports with the swing states going to Mitt. So far , this has not happened in any national poll or even state poll for that matter. It is very depressing. The taking lemmings seem to outnumber the makers in those swing states.