Charlie's just annoying because he's behind a paywall. Sabato is really awesome -- of the 'mainstream' pundits, he's probably the best out there -- and Nate Silver is pretty good too, though Silver is more interested in the mathematics than the politics and sometimes it can show.
Many Romney supporters are simply not interested in talking to pollsters.
As I previously coined it, The Obama Effect.
You actually know that silent majorities don't exist, don't you?
What reasons could have Romney supporters to lie in a massive way?
Occasionally, they do (two examples I can think of off the top of my head are the California gubernatorial election 1982 and the UK parliamentary election 1992); there's just doesn't seem to be any reason to think there is one in 2012. Though, they're called 'silent' for a reason.
I think Silver's analysis weights a bit too heavily in favor of victory based on small poll leads regardless of party. For example, when (pre-debate) polls had Romney leading by maybe a point or two on average, Silver was predicting a high 60's % chance of Romney victory. That said, I believe his calculated chances are based on who would likely win the election if held that day, not the chances that person will when on Election Day in a month, so the chances given are more understandable as they'll readily ebb and flow with changing poll numbers throughout the campaign.
EDIT: The pre-debate polls with Romney leading referred to above were for NC. Sorry.
