Rasmussen's Romney lead is simple to explain. They have this odd system of weighting by party ID where there's more Republicans than Democrats. The country almost always has a slight but clear Dem edge in party ID so naturally this shifts their results to make FOX news viewers happy.
He uses likely voters instead of registered voters. More expensive, but more accurate.
Not at this point. Given than there is no Democrat primary, Democrats aren't concerned by the election yet, so, they appear unmotivated and unlikely to vote.
A likely voter screen is only more accurate starting September.
I'll be honest in that I have not personally sifted through and compared Rasmussen to others and cross-referenced actual results, but typically he uses likely voters AND typically that is a BETTER method. It might not be right now, idk.
I think the rule of thumb is that Likely Voter models start to make sense around Labor Day, when most voters start to pay more attention.
