First big one, the poll is utterly meaningless, because of this at the end:

"The survey of 737 registered voters, taken Tuesday through Saturday, has a margin of error of plus or minus 5 percentage points."

1) Registered voters is a better gage then 'people', but republicans simply vote with higher frequency then independents or democrats.

2) It's not really a 5% error, because if you're going to look at 12 'battleground states' its only logical to look at them individually. For instance, did Santorum pick up Pennsylvania, his home state? That would be a big deal. Of course, had they sampled 737 votes broken equally by 12 states we'd have about 60 people per state, and the margins on that are huge.

So this isn't really a useful poll, it's just Gallup and USA today wanting a quick cheap one someone can right an article about. I also kinda find it silly to run polling on anything but specific upcoming states mid-primary because they know, like I know, that a single thin victory in one state can flip 5 or 10 points in the next one... also Gallup's polls were horribly off in 2008 primaries so I don't really trust them. Just for perspective, here was the polling this time four years ago, RCP Average

Poll Date McCain Romney Huckabee Giuliani Paul Thompson Spread
RCP Average 1/29/2008 26.7 20.0 18.7 13.3 5.2 -- McCain +6.7

And that was just when McCain was starting to pick up steam and Thompson had dropped out.