Accuracy of Senate Polls in 2004

[Thorgrim]

Here are all the "vulnerable" races in 2004:

SD
Poll Average: +1.3 Thune
Result: +2 Thune

SC
Poll Average: +9 DeMint
Result: +10 DeMint

NC
PA: Last poll after Oct 26: Burr +5
R: Burr +5

OK
PA: Coburn by 9 (and surging)
R: Coburn by 12
(Reason: Coburn surged the last week as voters started going more and more for Coburn and once you get to 10+ its hard to get exact results)

CO
PA: Salazar by 4.5
R: Salazar by 4

Florida:
PA: Castor by 0.5
R: Opponent by 2
(Reason: Surge in Bush support in last day caused coat tails the pollsters didn't factor in because it wasn't showing up)

Alaska:
PA: Basically a tie
R: Republican by 3
(Reason: no one spent serious money for public polls in Alaska)

And finally Kentucky had wildly different polls (one showing a tie, another showing a 15 pt lead...actual result was 2pts...not enough polling)
RealClear Politics - Polls
United States Senate elections, 2004 - Wikipedia, the free encyclopedia

Basically the Senate polls were AMAZINGLY ACCURATE in predicting a winner in every race less than 8-9+ average, the only exceptions were:

When the Candidate had less than a 1pt poll lead
or
There was a severe lack of polling

In 2006 There is NO LACK OF POLLING

So lets see the averages come Nov 6th and we should be able to predict the result

[motivez]

Interesting, but I guess the question is what about races like CT?

Can they really accurately poll those?

[Thorgrim]

Originally Posted by motivez

Interesting, but I guess the question is what about races like CT?

Can they really accurately poll those?

Remember Kentucky?

There are always exceptions

Just like Kentucky had a tie in one poll, +13,+15 in another...but the candidate was deranged, but it was a deep red state...

In Connecticut, I think about the polls that had Lamont within 2 points about 6 weeks ago, so it wouldn't be a COMPLETE SHOCK if Lieberman lost, I think everyone knows how messy this race is...

Ofcourse, note that 2004 all the close races were open seats basically, and there wasn't a national wave

This race involves incumbents, and a national wave...which means the poll is accurate but...the general rule is that an incumbent who can't get over 45-46% in a two way race is f00ked