Coakley still with a 90.6% chance to win. Normally, a lead of 8.2% would give Coakley a 97.5% chance to win. This is because only 7 of the 138 closest statewide general elections from 2004-2009 show a difference of 8.2% of greater from the final polling average to the final result (((7 / 138)/2) = 0.025).
However, because of the difficulties associated with special election polling, her chances of winning are only 90.6%. This is assuming average polling error in special elections to be the same as primary elections (7.0%), instead of general elections (3.9%). This makes a lead of 8.2% equivalent to a lead of 4.6%. In the 138 closest statewide general elections from 2004-2009, there were 26 instances where the final polling margin was 4.55% or more divergent from the final vote result (((26 / 138)/2) = 0.094).
What about those rumored polls? No rumored polls are included in my averages. Until those polls are released to the public, they will stay that way. This includes the rumor of a Republican poll showing Coakley up 11%, a Boston Herald poll showing Coakley up only 1%., and Coakley's internal polling showing her up only 5%. Let's look at each of these rumors:
Even if the 11% poll exists, it was taken too long ago to be included in the final averages, so it doesn't matter.
The 1% poll from the Boston Herald was first rumored 5 days ago. Media outlets don't sit on sponsored polls that long. It doesn't exist. It was either referring to the PPP poll that eventually showed Brown up 1%, or it was just bullshit.
Taegan Goddard's rumor about Coakley's internal polling only showing her up 5% must refer to the one-day sample of her internal polling on the 11th. This is because Coakley released an internal poll that was conducted from January 8th through the 10th, and Goddard reported the rumor on the evening of the 12th. At that time, the interviews for internal polling on the night of the 12th would still have been ongoing. So, even if Goddard is correct, he is referring to a one-day sample, which would have a high margin of error, and as such is not to be taken seriously.
In short, the rumored polls either don't exist, or they don't matter.
Why do the polls diverge so much? Which one is right? First, the polls don't actually diverge. As Mark Blumenthal showed on Sunday, they just project different turnout levels. All pollsters seem to agree that the higher the turnout, the larger Coakley's advantage becomes (and vice-versa). The campaign is tied among those who are "absolutely certain" to vote, but Coakley has wide leads among those who are less certain to vote.
The best bet, given the success of polling averages in predicting elections, is to just average all of the polls. With an average error of only 2.6% (actually 2.57%) from the final 15-day average to the final result, simple polling averages have proven to be the most accurate measure of election results available. It is more likely that there is a kernel of truth in all the polls than absolute accuracy in one or two of them.
Doesn't Rasmussen show the campaign tightening? My research shows, pretty conclusively, that including multiple polls from the same polling firm in the average, rather than just the most recent poll from each polling firm (which is what I did in 2008), reduces the error in the polling averages. So, the issue isn't if one polling firm shows the campaign tightening, but if the overall average is tightening. On that front, the Rasmussen poll did show the campaign tightening, but only from 9.8% to 8.2%.
What is the lesson in all of this for Democrats? This one is easy: the political environment is terrible for Democrats, and they are going to lose seats in 2010. Duh.
Why is Democratic turnout so low? This is very hard to say. No one has conducted a survey asking people who voted in 2008, but who do not intend to vote in 2010, why they don't intend to vote in 2010. Until such a poll is conducted, every theory about why Democratic turnout is down is just pure speculation. In most cases, pundits will just say that Democratic turnout is down because Democrats aren't doing what that pundit thinks they should do.
If Scott Brown wins, then the health care bill will not pass. There will be no 60th vote for Democrats in the Senate, meaning they have to go through Olympia Snowe. However, many members of Congress might well be scared off by the Scott Brown win, thus causing some lost votes on the right. Also, with everything the House Progressives feel like it has already had to swallow, some votes will probably be lost on the left, too. With only a three-vote margin in the House, a zero vote margin in the Senate, and the need to restart the month-long House-Senate negotiation process entirely, it is very difficult to envision the bill passing in any form if Scott Brown wins. In all likelihood, the whole thing unravels at the finish line.
With so much at stake, Dems, unions piling on ads. Given everything that is at stake in the Massachusetts special elections, both SEIU and the DSCC are making major advertising purchases in the campaign. They know Scott Brown is a longshot, but that he can still win. And they know what that means for the health care bill.
When is the election? The election takes place in six days, on Tuesday, the 19th.