Unlike Nevada, The Math In Not So Friendly in Iowa (UPDATED)

Friday, October 31, 2008

UPDATE: Make sure you read the update at the end...

So earlier in the week, I wondered why Obama was going to Iowa.

And Palin is going to Iowa, on Monday, the last full day of campaigning before the election.

I notice that Survey USA's poll that puts Obama ahead by 15 percent has a sample that is 29 percent Republican, 45 percent Democrat, and 25 percent independent. Now, I know the state GOP got obliterated in 2006. But in 2004, the exit polls put the state's electorate at 36 percent Republican, 34 percent Democrat, and 30 percent independent. Is this state really going to shift from a +2 advantage for Republcans to a +16 advantage for Democrats in a four year span?

Notice also this wrinkle in the SurveyUSA poll:

Among the 32% of respondents who tell SurveyUSA they have already cast ballots, Obama leads by 40 points; Obama leads by 3 points among those who tell SurveyUSA they have not yet voted, but will.

The early voting supports that wild Democratic skew, because as of Wednesday, it was 48.3 percent Democrat, 28.6 percent Republican, 23.1 percent independent/other.

But here's the thing. The state has 2,111,809 registered voters. 400,829 have voted early as of Wednesday. There are about 1.7 million registered voters left out there, and Survey USA, with this Democrats-ahead-by-16-percent-sample, finds Obama leading by three percent. Hmm.

Let's say the early voters split 70-30 (in keeping with Survey USA's number). That gives Obama 280,580 votes, McCain 120,249 votes. (Yes, I know, this gives no votes to third parties.) That gives Obama a margin of 160,331.

They're talking record turnout in Iowa; their previous high in recent history was 80.5 percent in 1992. Let's say turnout is at 81 percent, for a total of 1,710,565 voters in Iowa. In order to make up Obama's margin of 160,000 or so, McCain needs the 1,309,736 remaining voters to split in his favor by a bit more than 56 percent to 44 percent. Could it happen? Sure, particularly if the remaining voters are disproportionately Republican, as so many Democrats have already voted early. But having said that... it's a tough, tough uphill climb.

I understand Rick Davis said on a conference call today that their internal polls show Iowa is tied. I have little doubt that when the votes are counted, the final tally will show a tighter race than Survey USA shows. But putting McCain ahead, even by the skin of his teeth, will take a Herculean effort.

UPDATE: Many, many readers point out that the SurveyUSA sample of early voters is tiny, and thus the 70-30 early voter split in favor of Obama is probably wildly overstated. As one reader puts it:

If we assume that the D's and R's have voted their party, that's 28.6%R and 48.3%D, with 23.1% Ind. How do we allocate the Indy vote? The Survey USA poll you cited has the Independent vote divided 49-43 for Obama, so let's spit the 23.1% more or less accordingly — 12.7% for Obama and 10.4% for McCain. That comes out to an overall 61-39 edge for Obama among the already voteds, or an advantage of 22%. 22% of 400,000 is 88,000, a much smaller deficit to overcome.

In other words, by this math, McCain needs the 1,309,736 remaining voters to split in his favor by about 53 percent to 47 percent - a much more manageable task. Not easy, but doable.

10/31 12:02 PM Share