[andy58-in-nh]

Sabato definitely overestimated the Kerry vote for President in '04 (had Florida leaning for Kerry, for example - Bush won by a significant amount). On the House, he was dead-on though (called 232R-203D which was exactly right), and had the Senate as 53R-46D, which was off by 1R - with one Independent.

I think what he is missing this year is the same thing that all pollsters are missing: their methods no longer provide reliable results, as vast swaths of Republican voters are not responsive to polling (or included in it), and yet turn out in far greater numbers than ever before on Election Day. We'll see.

[Trust but Verify]

I agree with your comments. The pollsters are going to have to build new models or go out of business. With people using cell phones as their only telecommunications device and the widespread use of Caller ID to screen out unwanted calls, their methodology is failing them.

[kabar]

The polls are more accurate during Presidential election years. Turnout is always the most difficult variable to calculate. In 2006, turnout will be the deciding factor.