I know. If McCain was really down by six nationally, he would be down double digits in both states.
It really comes down to turnout. Take today's Rasmussen's numbers as a frame of reference. If turnout is like it was in 2004, he is oversampling Democrats by 2 points and understampling Republicans by 4 points. What he reports as a +5 lead for Obama becomes at least a +1 lead for McCain (I say "at least" because slightly more Democrats will vote for McCain than Republicans for Obama). Even if turnout is like it was two years ago, what was a +5 lead for Obama shrinks to +2 -- within the margin of error.
If turnout is like 2006, McCain/Palin will almost certainly lose. If it’s like 2004, which I think is MUCH more likely after the Palin selection, the chance of the Obamessiah winning is fairly small.