American Research Group, Dec. 31 - Jan. 2 (Dec. 26-28 numbers in parentheses)
Clinton 34% (31)
Obama 25% (24)
Edwards 21% (24)
This poll assumes 83% of those that show up will be Democrats. In 2004, 81% of those that caucused were Democrats. The low number of Independents and Republicans showing up in their poll could explain the high numbers for Hillary. ARG's polls have traditionally polled Hillary a few points higher than every other poll out there, so we should take that into account too.
Des Moines Register, Dec. 27-30
Now, much has been made of this poll. Namely, it assumes 60% of the caucus goers will be first-timers. Is this realistic? 2004 already set record numbers in attendance, with 45% of them being new. Those 45% aren't new anymore. Is that pool of new people really there? If so, that would mean a turnout the likes of which are unheard of. But even beyond that, some interesting numbers emerge.
Clinton remains the favorite of the party faithful, with support from a third of self-described Democrats. However, Obama is the clear choice of caucusgoers who affiliate with neither the Democrat or Republican parties, with roughly 40 percent of them backing him in the survey.
The support from non-Democrats is significant because a whopping 40 percent (actually, 39%) of those planning to attend described themselves as independent and another 5 percent as Republican. Only registered Democrats can participate in the caucuses, although rules allow participants to change their party registration on their way in to the caucuses.
So the DMR poll suggests that 45% of Iowans that show up to the Democratic caucus... won't actually be Democrats. And that also means a WHOLE lot of Republicans will be showing up. But look what happens. He's getting 39% of the 45% of caucus goers who are not Democrats. He's getting only 27% support from actual Iowa Democrats, who make up 55% of the caucus pool, according to the DMR. That means...
0.39 × 0.45 = 17.55%
0.27 × 0.55 = 14.85%
Of Obama's 32% in the poll, MORE support is coming from non-Democrats than from actual Democrats! Is this realistic, and is this a good thing? desmoinesdem, who is an Edwards precinct captain, doesn't buy these numbers from the DMR, either the turnout numbers, or the percent of non-Democrats that are supposedly going to show up. But who knows, maybe they nailed it, and we're going to see historic turnout tonight.
But Mark Blumenthal has more on what they meant by "Democrat" and "Independent", suggesting maybe the numbers aren't as much out of line as we'd think. It has to do with independent-minded Democrats identifying themselves as "independent" when asked.
McClatchy/MSNBC (via polling firm Mason-Dixon), Dec. 26-28 (Dec. 3-6 in parentheses)
Edwards 24% (21)
Clinton 23% (27)
Obama 22% (25)
They also did a reallocation (.pdf) based on second choices of supporters of the second tier candidates, and found this.
This poll assumes about 20-25% of caucus goers will be first-timers.
The rest of the polls don't tell us the percentage of Democrats or first-time caucus goers in the poll.
CNN, Dec. 26-30 (Dec. 14-18 in parentheses)
Clinton 33% (30)
Obama 31% (28)
Edwards 22% (26)
Zogby has been doing continual 4-day rolling averages in the last week.
(h/t poblano for the graphic)
Research 2000, Dec. 26-27
Strategic Vision, Dec. 28-30 (Dec. 26-27 in parentheses)
Obama 32% (30)
Edwards 29% (28)
Clinton 27% (29)
Insider Advantage, Dec. 28-29 (Dec. 16-17 in parentheses)
Clinton 30% (24)
Edwards 29% (26)
Obama 22% (27)
Insider Advantage also factored in second choices. After that recalibration, they found this.
That's because in their poll, Edwards was the second choice of a whopping 62% of Iowans who supported the second tier candidates.
And finally, from DailyKos, not a poll, but the most concise summary of what the three top Democrats stand for.
Clinton = "Me"
Obama = "You"
Edwards = "Us"
So... what do we know? Um... it's a 3-way race. Any of the top 3 can win this thing.