What's going on?
Quinnipiac has the New York governor's race between the popular Democrat Andrew Cuomo and long-shot Tea Party Republican Carl Paladino as a six-point spread. SurveyUSA has them at nine points apart. And Siena is out today with Cuomo blowing Paladino out of the water by more than 20 points.
I'm submitting my amateurish analysis of the poll numbers until Nate Silver weighs in with the definitive explanation. (Silver wrinkled his nose at Quinnipiac for not including Rick Lazio as a gubernatorial candidate, since he will likely appear on the Conservative Party line, and wrote yesterday, "This race has gotten a lot closer. Probably not as tight as 6 points — but much closer, nevertheless."
One thing to keep in mind is the self-fulfilling prophecy of early, and even inaccurate, polling. If everyone reports on a poll saying the race is close, the slew of media stories about the poll could create the reality the poll said it was reflecting.
As Steve Kornacki of Salon told me, "I do wonder whether the publicity of yesterday's numbers created will end up giving Paladino a boost -- so that even if the Q numbers were way off, now that everyone's been told he has such great momentum, he might really be closing in."
Pollster Peter Feld notes two other important factors: there's really not been any paid advertising yet. That'll shift the numbers. And, seeing the closeness of the race, Democrats could be jolted into action. That could mobilize some voters who have taken the race for granted so far.
"If Cuomo is at 49 percent among likely voters but 57 percent among registered voters, it means a low turnout is projected, one that skews heavily Republican," Feld says. "This fits with the national perception of an 'enthusiasm gap,' "
Now, he said, "the Democrats could easily fire themselves up later in the campaign, if the messaging improves (to be fair, we haven't seen the impact of ad campaigns yet, neither nationally nor in the Cuomo vs. Paladino race), and if so, the disparity between polling with registered voters and likely voters will probably reduce itself."
There's also a question of methodology.
If you average the numbers from polls that used a live interviewer (Marist, Quinnpiac and Siena), Cuomo leads Paladino 54.3 - 25.4.
And in this set of polls, there are no undecided voters, which seems unlikely.
Now, if you average the numbers from polls that use automated phone calls to survey voters (Rasmussen, SurveyUSA), Cuomo lead shrinks slightly, 52.9 - 36.4.
But the number of undecided voters jumps to a not-insignificant 15 percent.
Whether undecideds break to Cuomo or Paladino is unclear. In some ways, Cuomo is the incumbent in this race. He ran for governor once before, he's the son of a former governor, and he was elected along with a slate of Democrats who took control of Albany four years ago.
But Cuomo is running against Albany and the legislature and has taken on members of the Democratic coalition (Working Families Party, labor unions, "the legislature," etc.). And there is a sense that Paladino will continue his verbal assault on just about everyone. And the more the public pays attention to what he's actually saying, there's a chance he'll offend more voters than he attracts.
If you average all the polls so far (Quinnpiac, Siena, Marist, SurveyUSA and Rasmussen) Cuomo leads Paladino 52.1 - 40.1, with 6.4 undecided.
The polls that have come out so far which use "likely voters" and not just "registered voters" show the race being closer.
Quinnipiac has the race at 49-43. SurveyUSA has the race 49-40. What catches my eye in these is Cuomo hovering below 50 percent. For a one-on-one race where he, in some ways, is like an incumbent, that's a dangerous position.
And if you're not totally lost with all those numbers…Real Clear Politics aggregates the polls and puts the race at a 10-point spread for Cuomo.
Tomorrow, Marist is expected to release their poll.