Politics & Policy

Waiting on Iowa

Can polls be predictive? Sure. Just not yet.

Venture a guess: What percent of Iowa’s registered voters took part in the 2004 Democratic caucus?

Iowa had about 2.2 million adults eligible to vote that year, and 124,331 participated in the Democratic caucus.

That means that a mere six percent took part.

As we see poll numbers coming out of Iowa this fall, we ought to keep in mind that the final preferences of those six percent or so on caucus night and the leaders in polls conducted in the closing months of the preceding year tend to be distant relations.

For example, on Halloween 2003, a poll conducted for KCCI TV-8 in Des Moines put Dick Gephardt and Howard Dean tied at 26-percent each. John Kerry was third with 15-percent, followed by John Edwards at 8 percent.

In October 1999, the same poll put George W. Bush 38 percent and Steve Forbes at 17 percent. That seems respectable at first glance, although the final results of those two were much closer, 41 percent to 30 percent. But that poll put Elizabeth Dole in third place at 13 percent, Gary Bauer at 9 percent, and a fifth-place tie between Pat Buchanan and Alan Keyes. A lot can change in three months: by January, Dole had dropped out of the race and endorsed Bush, Buchanan had left the GOP, and Keyes managed a respectable third-place finish at 14 percent.

But while we chuckle at past comments from pollsters calling Dean a “juggernaut” we should note that most pollsters’ results do get closer to the final results in the last days of the campaign. For example, the Iowa Poll, released the day before the caucuses in 2004, found that 26-percent of the 606 Iowans they polled were for Kerry, with Edwards at 23-percent, 20-percent for Dean, and 18-percent for Gephardt.

The final results were Kerry at 38-percent, Edwards at 32-percent, Dean at 18-percent, and 11-percent for Gephardt. Even if the percentages were off (by double digits for Kerry and Edwards!) they did manage to get the order right.

Pollsters are still accounting for new factors in this year’s caucus. “Figuring out who’s going to show up is something you deal with on every poll, but it’s just so much easier in a primary state,” says pollster Scott Rasmussen.

What’s making it even more complicated this year is that the caucus is on January 3. College students will not be back from vacation, so that’s a factor. The whole flow of campaigning will be different — how do you campaign between Christmas and New Year’s?

John McLaughlin, who is polling for Fred Thompson’s campaign as well as for the MTV/MySpace candidate forums, is on the ground in Iowa. He notes that the University of Iowa poll released Monday obtained its sample “from a random sample of 35,000 residential phone numbers across the state of Iowa. This list was purchased from an independent company.” Caucus-goers were self-identified.

McLaughlin expresses skepticism that residential phone numbers are sufficient; he suggests phone lists of previous caucus-goers, or at least primary voters from 2006, would be the better place to start.

“You have to have not just a better methodology, but a better sense of people’s behavior,” McLaughlin says. “There’s no sense in polling people who have no history of voting before in a Republican primary… If you don’t vote on a Republican primary, why would you attend a caucus? That would be extraordinary.”

Does this mean that pollsters expect no first-time voters? Not quite. McLaughlin says that on the margins, certain candidates do bring in first-time voters from demographics they have a particular appeal:

Romney may bring in some Mormon voters, Huckabee could bring in Baptists…Fred is bringing in gun owners, some anti-tax people. But the voters they attract don’t come out of thin air; they come, by and large, from people who have a history of some sort of Republican or conservative activism.

Keeping in mind that McLaughlin works for Thompson, here’s his read on the caucus environment from on the ground: “Giuliani and McCain, I don’t know where they’re going to play in this caucus,” he says.

Fred’s had movement and he’s doing pretty well. Huckabee’s doing pretty well. Romney still the frontrunner, but I think he’s more vulnerable than people perceive. There’s still some volatility and fluidity, but the activity is all on the right, all conservative. On January 3, the people who are coming will be coming for the conservatives – Thompson, Huckabee and Romney.

Rasmussen sees a similar dynamic, with a bit more good news for the Romney camp.

“We looked at a variety of turnout models and while the numbers shift a little you see the same general pattern,” with Romney ahead, and a tight bunch consisting of Thompson, Huckabee, and sometimes Giuliani competing for second place, McLaughlin says.

We couldn’t come up with a turnout model that showed Romney behind. When we had the tightest voter-screen model that you could come up with, we found Huckabee a little better than Thompson – but the numbers were so small, I don’t want to draw too many conclusions from that.

Is that tightest-voter-screen model giving us a sense of what might happen if, say, a blizzard blanketed the state on the night of January 3?

“I guess it could be,” Rasmussen says. “In Iowa, there is a commitment factor you don’t see quite the same anywhere else.”

“The history of Iowa suggests you’ll see some motion in the numbers in the final months,” McLaughlin says. “Bush fell from 50-to 41-percent; he wasn’t in a position where he’s going to lose his lead, but he lost ground.” He noted that in 1996 there were significant shifts in the last 12 days – Dole gained, and Lamar Alexander and Pat Buchanan rose and passed Steve Forbes. “It’s not so much that the polls we’re seeing right now are inaccurate. It’s that you could see motion within the voters.”

– Jim Geraghty writes the Campaign Spot blog on NRO.


The Latest