A note about Iowa polling

[Kurt_and_Hunter]

(I had to wait until a few Clinton-leading Iowa polls in a row came out, because I don't want this observation about the polls to be tainted as special pleading based on one result or another. Right now, there is no clear Iowa leader, so it's safe to say this.)

All Iowa polls use likely caucus-goer models that include 200%-400% of the number of people anyone actually thinks will caucus. Likely models always include more people than will actually vote because people tend to overestimate their odds of voting. Likely caucus-goer models are even more askew. In practice, very few people caucus as a percentage of the electorate.

So we know that the caucus will be decided entirely by which people polled as likeliest don't end up caucusing. Barring a big break-out by one of the top tiers, it is a turn-out game now. Every poll coming out now could show any of the top three ahead by making different likely caucus-goer assumptions that are equally valid. (One poll noted that their result could be anything between Obama up 10 and Clinton up 6, depending on which turn-out model they chose to use.)

So in a race this tight (at the moment) the polls cannot predict the winner. Turn out effects swamp public opinion effects.

The polls are, however, valuable for picking up relative movement... a particularly important measure because momentum is a very real factor in all elections, and particularly in primaries. If Biden surged to 15% in a couple of late polls it would be likely that he would exceed 20% in the actual caucuses. If Clinton drops 10% in several late polls it suggests that she is unlikely to win, and so on.

[gateley]

1. What I've found interesting, since I'm paying attention this year, is that a large

number of people DON'T caucus.

From being on DU, and seeing how enthusiastic we all are, plus all the Iowa press coverage, I assumed every single voter in Iowa was chomping at the bit to get out and participate. When I heard the actual percentage of those who do caucus (I don't remember what it was, now) I was surprised.

[Kurt_and_Hunter]

5. I was looking at a Research2K poll subsets, and 38% of their sample was independents.

That made me wonder what percentage of Iowa independents have ever caucused. 38% seems like a hell of a lot to have in a likely caucus-goer model, since I think you have to change your registration at the caucus to the party you're caucusing with. (If that's inaccurate, somebody please correct.)

[frazzled]

9. Only 6 to 10% of eligible Iowa voters actually caucus

Amazing isn't it, after all the hooplah.

If this is a typical election, somewhere between 6 and 10 percent of voting-eligible Iowans will bother to show up to a caucus. Yes, you read that right. Those vaunted Iowa voters are so concerned about the issues, so involved in the political process, so serious about their solemn deliberative responsibilities as guardians of the first-in-the-nation contest, that nine out of ten can't manage to haul their butts down to the junior high on caucus night.

http://www.prospect.org/cs/articles?article=the_myth_of_the_rational_iowa_voter

I can't seem to find a statistic on how many of these are independents (yes, they can change their registration at the caucus itself, but it would seem unlikely many do).

[bunnies]

2. right on.

which is also why NH polls dont mean anything at this point either. Iowa winners will get big bumps in NH. And as of yet, theres no way of knowing who the Iowa winners will be.

[suston96]

3. Good post. I agree.

Many observers honest enough to admit it are saying the same thing.

This Iowa race, Democrat or Republican cannot be predicted. How about guessed? Sure. Take your best guess.

Here is mine - because of the 15% rule anything can happen. My guess - my wish - is that it turns into a dead tie so that this undemocratic process in Iowa ends its relevance forever.

[durrrty libby]

7. Ha That is a good wish. The candidates spend months and millions wining and dining

the state and only a few even bother to turn out

The system is a ridiculous archaic ritual. It needs to go

[NewYorkerfromMass]

4. Hillary is a given factor because of GOTV machine

when you have the party apparatus behind you, as Kerry did in 2004, your voters are getting to the polls no matter what.
And the unions are very important in this regard. Hillary has a lot of union support, but Edwards and Obama have also aligned some.
I think all things considered, its still Hillary's to lose. The worst outcome I forsee for her is a 3 way tie.
Of course NH becomes ground zero at that point.

[Kurt_and_Hunter]

6. There are other weird factors in play that favor Edwards.

He has a very high percentage of repeat caucusers in his base, which I think makes them more reliable than folks who say they are going to caucus for the first time this year.

And Rural caucus sites get weighted disproportionately high. Edwards is strongest in rural areas, Hillary and Obama more in city and suburb.

It's all full of on the ground nuance that we can only guess at. (Unless we are on the ground in Iowa)

[Quixote1818]

8. Also, the young vote tends to flake out and not show up

They want to sound mature and tell pollsters they will show up and vote but then they end up at the bar or in this years case, home for Christmas.

Obama could get hurt by the flaky young voters because that is where he is strongest.

[hayu_lol]

10. Quixote...your 2nd sentence explains...

Obama's going for the bus trip or Iowa college students from Illinois back to Iowa. Get 'em on a bus, take them there, and then take them back home.

Smart move on his part. Looks like only Hillary caught what was behind Obama's move.