Ian Lazaran over at C4P has done a great job of deconstructing this statement. He points out that this was the same poll (Selzer Poll) that had Branstad up by 28 points over Vander Plaats when his actual margin was only 10. Bottom line: this usually-accurate state poll badly oversampled moderates which favors Romney and to some extent Gingrich.
The author of this AP article, of course, fails to point this out. Hard to imagine Palin trails Romney in favorability in Iowa when she leads him (and the other wannabes) nationwide among Republicans. If Huckabee doesn't run and Palin does, she should win Iowa easily.
I am sick to death of moderates, whatever the hell they are.