News from the Votemaster
The Michigan primary is on Tuesday, and the pollsters are at it again, hoping there will not be any last minute crying by any of the candidates to foul up the polls. There is no race on the Democratic side (because Hillary Clinton is the only major candidate on the ballot) but there is a real battle on the Republican side between John McCain and Mitt Romney. The five new polls average out as follows:Mitt Romney: 27.2%
John McCain: 26.0%
Mike Huckabee: 16.8%
Fred Thompson: 5.8%
Rudy Giuiani: 5.4%
Romney really ** ought ** to win this one. Michigan is his home state and his father used to be governor. If he loses to McCain here, it will be tough to explain. If he pulls this one out of the fire, he'll live to fight another day.
We also have a new SurveyUSA poll of Florida, where Hillary Clinton has a 19 point lead over Barack Obama, 51% to 32%, with John Edwards trailing at 11%. On the Republican side, John McCain leads Rudy Giuliani 27% to 19%, with Romney and Huckabee tied at 17% each. This is a must win state for Giuliani. He skipped all the earlier contests to focus on the only southern state mostly populated by transplanted New Yorkers (at least during the winter). It is do or die for him.
Discussions of why the polls were off in New Hampshire keep going on. A lot of people are claiming it was the "Bradley effect"--white voters lying to the pollsters to hide their racism. That just doesn't hold water. Take a look at the average of the last 15 polls (all ending Jan. 5 or later) compared to the election results:
The polls hit Obama's percentage nearly spot on. They were off by 0.3%, well within the margin of error of 3-4% for this kind of poll. Notice that the sum of the polls is 85.3%, meaning that about 15% were either undecided or for one of the second-tier candidates. What happened on election day was a small loss for Edwards and a big break of the undecideds for Hillary Clinton. The people who told the pollsters they would vote for Barack Obama by-and-large did so. There is no evidence of a "Bradley effect" in this data.
-- The Votemaster