Obama wins New Hampshire primary in a landslide!
That was the prediction of the University of New Hampshire Survey Center and most other pollsters prior to the primary. In fact, Hillary Clinton won the primary, causing much hand wringing. This shows us that we need to be cautious with polling data. It is just part of the picture.
The UNH poll released this morning by The Boston Globe, showing a safe 15% lead by Coakley among likely voters, does not ring true to me. UNH started polling on January 2 and finished January 6. Rasmussen, which polled during that same time period, showed Coakley up by 9% among likely voters but only 2% among definite voters. PPP polled more recently and showed Brown up by 1%. The Boston Herald is to release a poll which reportedly will show Coakley with a 1% lead among likely voters. [Note added: Jules Crittenden disputes that there is a pending Boston Herald poll.]
The UNH-Globe poll is an outlier, by far. The three other polls show this as a single digit race among likely voters, with Brown’s voters far more highly motivated. [Added] Even the UNH-Globe polls shows that among the voters who are listed as “extremely interested in election” it is an even race, at 47% each.
All the polls could be wrong, as they were in New Hampshire, but we have what we have, and it appears to show a close race with a small advantage to Coakley.
I doubt many people, including Democratic political operatives, believe the UNH-Globe numbers of a sweeping Coakley lead reflect reality on the ground. I certainly don’t.
This is a tight race. Brown and his supporters need to finish strong to win.
Update: TPM has a reasonable assessment of how the UNH-Globe poll could differ so widely from the PPP poll:
So what explains this crazy spread? One thing to note is that the PPP poll is a bit more recent, though seemingly not enough to explain the huge spread. At least not all of it. Another interesting thing about the two polls is that they’re not that far off on Coakley’s number: PPP has her at 47% and the Globe has her at 53%. The difference is in Brown’s number — 48% vs. 36%. As I said, I think the whole story here is that screen the two pollsters are using to see who’s is going to vote.
Now PPP is actually a Democrat-associated firm. So while I think I think they’re top-notch, no one who’s inclined to be suspicious should have any thought that they’re somehow biased in favor of Republican candidates.
How have they done recently? Well, we have two races to look at. PPP came out with a raft of polls just before the November special election and noted that they were seeing a much more Republican electorate than other pollsters were. They picked up the fact that Christie was starting to surge in the last couple days in New Jersey — they had Christie up by 6 points. But they also had Doug Hoffman clobbering Bill Owens by 17 points in their final poll, even though Owens won.
All of which tells me there’s some reason to think that PPP is oversampling Republican voters. But I wouldn’t bet money on it. If Dems won’t avoid a very, very bad day on January 19th, they need to really get in gear on this race.
I particularly agree with that last sentence, which is true for both parties. Brown can win this, but only by continuing his surge.
And HotAir has a good post which hits on the same sampling and screening issues, What’s up with the Mass. Senate polls?
And Nate Silver’s analysis “any of them could be right.”