Habla la gurú de las encuestas de Iowa

The Atlantic entrevista a J. Ann Selzer, que lleva un cuarto de siglo haciendo encuestas del Caucus de Iowa para el Des Moines Register.

When did you start polling the Iowa caucuses?

I came to the Des Moines Register in December of 1987. I was on their staff when it had a complete polling unit inside the newspaper. The '88 caucuses were in that February, and they had put in place a technique of trying to save some money by recontacting people from previous polls. And I said, well, let's take a look. We have fresh numbers here, we have recontacts here. We were publishing that George Bush was going to win the Republican caucuses, and I went to the editors and I said, 'I don't think that's true. I think Bob Dole is going to win.' They said, 'What?' So I showed them the data, and they said, 'What would this take to fix?' I said, 'Money.' They said, 'OK.' So I very early had a chance to prove my credibility, because Bob Dole did win the caucuses.

I understood the urge to do the recontacting, because finding likely caucus-goers is very difficult. There are not many of them. But for whatever reason that I still don't know, [the recontacts] were more inclined to be George Bush supporters. We didn't have an explanation for it, but it didn't matter -- there was distortion.

Were there other times when your data picked up things people weren't seeing?

In 2004, Kerry had just been sitting -- he was kind of like the Romney of 2004, his numbers wouldn't move. And then in our last poll we showed that surge, showed John Edwards [up], showed Howard Dean [down]. And I still keep this graph, because everybody thought Howard Dean would win. I call it the 'Howard Dean graph of doom' because he was plummeting in the final days. Edwards, if the caucus had been two or three more days out, he could have surpassed Kerry. He was the one that was on fire.

(...) Has there been a time you missed the mark?

With the caucuses, we've been very fortunate. In 2004, we had Kerry taking Iowa [in the general election] and it killed me, it just killed me. [George W. Bush won the state by less than a percentage point.] Des Moines University was my client and [current and former Iowa Gov.] Terry Branstad was president at the time, and he happened to walk by a room where I was meeting with one of his vice presidents the Thursday after the election. And he came in and he said, 'You couldn't have gotten it. You finished interviewing Friday night. The things that happened Saturday at that [Bush] rally in Sioux City, that changed the vote in the 5th District, and you couldn't have gotten it.' It took another six months for me to go, OK, I couldn't have gotten it. It still feels terrible.

Things happen at the very end. We have to stop polling at some point. We'll finish polling this time around Dec. 30, because Dec. 31 they're going to write the story for publication on the 1st. There's still two more days until the caucuses. So there's a lag time when things will still be happening -- I worry about that. I'm a big worrier, can you tell?

What happened in 2008, and what can we learn from it to apply to this year's caucuses?

In 2008, our final poll came out New Year's Eve night. Within half an hour both the Edwards campaign, but more vociferously the Hillary Clinton campaign, had memos out saying, 'Pay no attention to this poll. She's assumed this, she's assumed that, this is crazy, her turnout model is unlike anything we've ever seen before, she should be fired.'

I had Obama by 8 points. We had had Obama in the previous poll. But this poll also said that 60 percent were going to be first-time caucus attenders. That was high by historic standards. Forty percent were going to be independents -- that's high by historic precedent. And that afternoon one of the key people in the Clinton campaign called me up, a friend of mine, and said, 'I've always trusted your polls -- until now. I've knocked on 99 doors and I don't find this lurking Obama support.' And I said, well, tell me about the doors you've knocked on. 'Previous caucus attenders and registered Democrats.' Well, you're not going to find it there.

And so flash forward to now: What an interesting lesson to learn about Iowa. The caucuses will be what the candidates make of it.

[This year,] the national media and I think some of the candidates want Iowa to just replay what happened in 2008. Some of the candidates are in this race because they thought they could be the next Mike Huckabee. I think that's Rick Santorum, I think that's Michele Bachmann, I think that's Rick Perry to a certain extent. And I think they've misread Iowans in thinking that there would be that holdover wish for that kind of candidate. Really, times have changed, things have moved on. So I think you end up with candidates who aren't resonating because they're not talking about fiscal ideas to solve the economic problem. They're focusing on the social ideas they think Iowa caucus-goers would spark to. You have Romney not wanting to play here, you have Jon Huntsman saying he's not going to play here -- people who are potentially on the fiscal side who think they can't win in Iowa. And the data just say that's not true.

Huckabee won because he was really the only one -- he was the social conservative, and the rest fragmented the field, so he stood alone. Well, now you have all of these social conservatives who are fragmenting the field. Romney, Huntsman, had they decided to play, [could have had a good chance by mobilizing moderates].

What's different about the campaign this year?

What's different is -- and maybe this will change in the next month -- [in past cycles] you had this feeling of candidates here, kind of a pulse of their supporters, lots of public events. You felt things happening. I haven't felt that at all [this year]. You look at the number of candidate days here, it's really small. The way that they're campaigning is very different. We have all of these debates that have been very high-profile. It's a different campaign.

Only Rick Santorum is campaigning the way we would expect people to campaign. He's been to all 99 counties. He has increased his staff. He does all of these public events. It's not working for him, and you obviously have to have the right candidate for those techniques to work. But it's not working, and I think [that's] because people are really wishing for something else. People talk about these [last few elections] have been wave elections and I go, well, it's not that the electorate is swinging back and forth in such a partisan way. It's that they say, 'Not this,' 'not this,' and 'not this.' And that's why Herman Cain I think is so appealing. He's Mr. Not This.

In our Bloomberg poll we had an analysis of how many people had been contacted by each of the campaigns. Ron Paul was first, followed by Michele Bachmann. And the secondary analysis was to say, OK, if you've been touched by that campaign, who's your first choice? So we could kind of look and see the effectiveness of those touches. Santorum goes from 3 percent to 6 percent among people his campaign has touched, and that's double, but if you're a small number it's easy to double it.

Michele Bachmann gets a one-point lift [among voters her campaign has contacted]. It's not doing her any good. Who gets the lift is Gingrich. His campaign contact number is high 20s, low 30 percent. But he gets 32 percent first-choice votes among people his campaign has contacted. That's almost double the 17 percent he gets overall in the poll. That number is a very strong number for him. What [voters] have seen of him they liked, and what they have seen of other candidates didn't impress.