6 problems with the Iowa polls
Trump and Clinton are leading. But maybe they're not.
By
Steven Shepard 01/31/16
Donald Trump and Hillary Clinton enter Monday with small, but consistent, leads in polls conducted in the days leading up to the Iowa caucuses. But what if the public polls are wrong?
What if Iowa Republicans choose Ted Cruz or Marco Rubio over Trump on Monday night? What if a surge of Bernie Sanders supporters propels the Vermont senator over Clinton in the Democratic caucuses?
After a series of high-profile misses in recent elections, America's pollsters outlined the hazards of polling the 2016 Iowa caucuses in interviews with POLITICO over the past week. Here are six things that complicate their efforts and could lead to an official result that doesn't look much like the one the polls appear to predict now.
1. It’s a caucus – not a primary.
Participating in a caucus is harder than voting in a primary election. Unlike primaries, caucus voters can only cast their ballots in the evening, at a designated time. And the process is far more time consuming than arriving at the local polling place and pulling the lever. That has implications for turnout.
But what really complicates things for pollsters is the very nature of the caucuses, where voters can and sometime do change their minds during the event. While the process is different for each party, both Democratic and Republican caucuses are communal experiences. Most states don’t allow “electioneering” within polling places, but the caucuses encourage it. Supporters give speeches, making the case for their candidates. In the Democratic caucuses, voters backing candidates who fail to meet a viability threshold in an initial vote – usually 15 percent – are lobbied to pick another candidate in the second round of voting.
Pollsters measure the intentions of voters in the days leading up to the caucuses. Even the entrance polls, which will be cited on all the cable-news networks Monday night, ask caucus-goers their vote preferences when they arrive. But none of that accounts for people who switch to another candidate during the caucus.
J. Ann Selzer, the long-time pollster behind the storied Des Moines Register survey, said last week that lobbying does have an impact.
“You may be someone who is for John Kasich, and you get approached by 17 Jeb Bush people” at the caucus site, she said. “All a poll can do is estimate what people intend to do,” Selzer said. “But it’s a process designed for people to change their minds in the room.”
2. Who’s in the poll?
Pollsters call the caucuses low-incidence events. Attendance at the 2012 GOP caucus was only 6 percent of the overall pool of registered voters in the state. Four out of every five registered Republicans didn’t participate four years ago.
But when pollsters call voters about their intentions before an election, those voter lie – or, more charitably, overestimate their own plans to vote. So pollsters are faced with the challenge of weeding out those who won’t actually show up on Caucus Night.
It’s a decision they often make before making a single phone call. Some pollsters choose to dial phone numbers in the state at random, so that every Iowan has an equal chance of being contacted. But other pollsters argue that that sampling frame is too broad. They restrict their surveys to those with a history of past turnout in caucuses, primaries and general elections.
More active voters are more honest about their own chances to vote than those who have a spottier voting record, according to Patrick Murray, the director of the Monmouth University Polling Institute. “People who have a higher proclivity of voting – if they say they are more likely to show up this time, they are more likely to show up this time. If they say they are less likely to show up this time, they are less likely to show up this time,” Murray said. “They are less likely to lie.”
There’s a middle ground, including how Selzer conducts polls for the Register and Bloomberg Politics: Call all registered voters in an effort to exclude non-voters – but not to cut out those who might be more motivated this year than in the past.
3. Trump and Sanders might change voter profile.
Getting that right mix of voters is complicated in many ways by Trump and Sanders, who are appealing to non-traditional caucus-goers: Trump has a 16-point lead among respondents to the Des Moines Register/Bloomberg poll who say they are participating in their first Republican caucus, while Sanders leads first-time Democratic caucus-goers by 19 points.
That’s why it’s widely thought that a surge in turnout would boost those two men – and hurt Ted Cruz and Hillary Clinton, respectively. But will those people actually show up on Caucus Night?
Of course, pre-election polls can’t determine that. But some are already arguing that the polls are including too many voters who won’t turn out on Monday, usually using some back-of-the-envelope math.
Jeff Roe, Cruz’s campaign manager, tweeted Saturday night after the release of the Register survey that Selzer’s poll was making assumptions about turnout that were unlikely to be borne out. Roe pointed out that the poll interviewed 3,019 registered voters – 602 of whom were determined to be likely to attend a Republican caucus.
Roe’s
tweet: “602/3019 GOP voters is 19.9% of the 1.9 active voters in the state; simple math says: 386,000 turnout for GOP
#youhadonejob.”
But all registered voters isn’t necessarily equivalent to those registered voters who would participate in a poll. Studies show that voters who are more likely to take the time to answer pollsters’ questions are also more likely to turn out for an election.
For her part, Selzer said her poll isn’t intended to be a turnout estimate. “I don’t think polling is a good tool for measuring turnout,” Selzer said in a separate interview late last year. “Polling is an estimation tool. So if we think there’s going to be a difference of 20,000 voters out of a state of 2.1 million, you’re dealing with such small fractions of percentage points that would make a difference.”
4. Turnout's king but it's hard to predict.
While the polls themselves aren’t necessarily indicative of what turnout will be on Monday night, turnout will affect the post-game verdict on whether those polls successfully predicted the outcome in Iowa.
Monmouth University polling data released last week looked at the horse-race at various turnout levels and showed
Trump and
Sanders both performing best with greater numbers of voters participating. Lower overall turnout, on the other hand, would appear to benefit Cruz and Clinton, respectively.
In the age of big-data campaigning, each candidate’s team knows which turnout levels are helpful or unhelpful for them – and they are acting accordingly, trying to motivate voters more likely to support them on Monday night.
But there are a number of other factors, too. There are the actions of the other candidates, and then there are external dynamics, like a snowstorm expected across much of Iowa on Tuesday that could begin Monday night.
5. Democrats count delegates, not votes.
The polls of the Democratic caucuses could be wrong – either Clinton or Sanders could romp to a decisive victory. But that will be harder to evaluate than on the Republican side.
That’s because, while Republicans report the total number of votes tallied by each candidate, Democrats only release the number of delegates won by each candidate at every precinct.
It’s possible to estimate vote totals by multiplying the percentage of delegates won by the candidates by the total number of attendees, but that’s imprecise. And, in many precincts, it’s likely any supporters of former Maryland Gov. Martin O’Malley will be up for grabs, as O’Malley appears unlikely to meet the viability thresholds in the initial balloting.
“We are predicting something,” said Monmouth’s Murray, “where we’ll never know if we’re right or wrong.”
6. Polling is harder now.
Recent years
haven’t been kind to political pollsters. And continuing technological changes in the ways in which Americans communicate mean 2016 could be even more difficult than other elections in which the polls were wrong.
That means – in addition to trying to thread the needle and accurately predict the volatile caucuses – pollsters have to grapple with all these changes.
Asked by POLITICO last fall if polling this election was more difficult than in the past, Selzer said: “Yes, but I would have said that four, eight, 12 years ago. At every given point it’s more difficult than it was in the good ole’ days of George Gallup.”
We Asked Iowa’s Top Journalists What’s Going to Happen on Monday
Spoiler: They have no idea, either.
By POLITICO Magazine 1/29/2016
What’s really going on in Iowa? Will masses turn out to caucus for Trump? Is Bernie Sanders going to be able to ride his momentum to a shocking win?
Instead of speculating, we went straight to some of the people who know the situation on the ground best: The Iowa political editors and reporters who’ve been watching the race up close, and have lived through one cycle after the next.
In a roundtable interview moderated by Politico senior politics editor Charlie Mahtesian, they offered some surprise insights: Voter turnout might not be an exception this year;
Hillary Clinton’s impressive ground game has done a solid job countering the Bernie Sanders wave; and Donald Trump—despite the polls, despite the crowds and despite the clear on-the-ground excitement—could still suffer an embarrassing defeat, not just to Ted Cruz but also to Marco Rubio.
Behind it all lies the weariness of a state that has been uniquely inundated by the American campaign machine this season. After all the YouTube ads, the mail, the attacks on TV, jokes one editor: “I can’t wait to see a fertilizer commercial.”
***
Charlie Mahtesian,Politico:
Can you tell us a little bit about what the national media is missing about this story? What is it that we don’t understand about the state of the election or about Iowa, or is there something we consistently get wrong?
Ed Tibbetts, Quad City Times: My perception is fairly positive when it comes to coverage. Over the years I was disabused of the notion that folks just parachute in. Yes, there are some who come for a brief period of time, but I think that a lot of the national media come and stay awhile, and it shows.
Christinia Crippes, Waterloo-Cedar Falls Courier: I sometimes feel like it gets missed that people are really not a hundred percent committed. People really do make up their minds at the last minute, and I think sometimes it kind of comes across that things are settled, or that some particular person is going to do well, when in reality these people are going to see multiple candidates. You see the same people on stops, and even if they say at one event that they like this person, they could go to the next one and say, ‘You know what? I was really impressed with this person, too.’
CM: How does this compare with prior elections? Is the excitement level about the same? Is it heightened?
Dale Alison, Hawk Eye: I’ve been here since ’85, and I think that it’s really a tough one to answer. The Donald Trump folks—those guys have been committed from Day One, and they have tended to draw in more and more. They’ve been good recruiters.
On the Democratic side, I really think Hillary Clinton has some rock-solid supporters, and she’s got her really solid core. But the enthusiasm is clearly with Sanders. And poor Martin O’Malley is just kind of left out.
Amalie Nash, Des Moines Register: This is my first experience with Iowa caucuses. I’ve asked that question of a lot of people here, both on the [
Des Moines Register] staff as well as out in the community, and I think because of how large the field is, there has been a lot more interest from the beginning, and then when you have someone like Donald Trump running, that sort of ratchets up the coverage. I think people are a little bit more interested and excited.
We have a project that focuses on getting millennials involved with elections. And so, as we’ve been doing that and talking to a lot of millennials and having them write for us and do various things, I’ve been impressed with how engaged they are. I think part of that has to do with Sanders, because a lot of them are leaning in that direction. He’s got that young support in a lot of different factions here. It definitely seems like there’s a lot of interest from all different sort of demographics out there right now.
Bret Hayworth, Sioux City Journal: The
Sioux City Journal covers 15 of Iowa’s 99 counties—and my day yesterday was spent calling county auditors and getting voter registration numbers, and we compared voter registration this year one week before the caucuses to the 2008 and 2012 elections. Very, very similar numbers.
AN: Our pollster keeps the voter list updated, and so she’s constantly looking at that and sent us an update on January 18, saying that at the end of the prior week they had about 10,000 new registrants on that list, divided pretty evenly between male and female.
ET: It’s difficult to figure out from one caucus cycle to the next whether the electorate is more excited. In 2008, we saw a great level of interest and excitement that I would say is probably similar to what we’re seeing in this cycle. What strikes me as being different, as it is in every cycle, is just the amplification of media and avenues for messaging.
DA: I think you can draw some parallels between Bernie Sanders in 2015 and Barack Obama in 2007. And I think there was that constant support for Hillary Clinton, even in that cycle, but the people who could be influenced tended to drift to Obama back then, and I think those same people are probably drifting to Sanders.
On the Republican side, I think everybody can agree that there has never been anything quite like Donald Trump, and everything seems to be reflected through that lens. He’s consumed everything on the Republican side.
CM: Will Sanders and Trump supporters actually turn out? What can you tell us about what the electorate is going to look like Monday? Do you think the energy surrounding the Sanders and Trump campaigns will actually translate into turnout in the way it did with Barack Obama when the same questions had surfaced?
DA: Absolutely. I look at Sanders and Trump as generating the same type of people who are dissatisfied with the status quo, who are dissatisfied with the way things are going. Those people have rallied around somebody who is different, and I think those people are highly motivated to turn out Monday.
CC: I live next to a college town and so from my standpoint, I see better that Bernie Sanders' people are organized and doing events that teach people how to caucus. They had an event the other night that explained to people the process of caucusing. At least in this part of Iowa, I see that Bernie Sanders’ people will probably be organized and show up.
DA: It would not surprise me in the least if Donald Trump comes out of this blanked. Those [voters interested in Trump] may show up for the Republican caucus, but they may turn out voting for somebody else, and I could see the Trump campaign essentially vanishing after this.
ET: Some of the strategists that I’ve talked to say that at least on the Democratic side, they don’t see a turnout like we saw in 2008. On the Republican side, some of the smarter folks I’ve talked to have talked about the possibility of a larger turnout. Given just how small a slice of the electorate the caucus participants are, it’s really something that’s just so difficult to know.
CM: The Iowans I’ve talked to are expecting a record-breaking turnout this year. Are there any contrarians among you who think that maybe that’s not the case, that maybe the level of excitement among Republicans might be overstated, that this might just turn out to be a standard year?
BH: I go simply on interviews that I’ve done with people, but I think it’s perhaps overstated that it’s going to be some record turnout amongst Republicans. Talking with some county officials again, as I’ve done in the last few days, I talked to one who—in the largest county up here in northwest Iowa—said it looks like it’s a typical presidential year.
DA: In this state, I think it’s going to be highly dependent upon the weather. If it looks like it’s snowy or rainy, which I think is a possibility, that could have a huge impact.
CM: We’ve talked a lot about the rallies and the field organization and the mechanics of the campaigns. I was wondering if anyone could go out on a limb and about who you think has run the best campaign in Iowa. Which stands out as a campaign that’s really wired in to the state, organized everywhere, and has attracted the notice of Iowa?
DA: I think Hillary Clinton has maintained the support she had in ’08, and then she went out and purposely tried to round up the Obama supporters, and I think she was pretty successful at that. But I don't know that anybody saw the Sanders wave coming, and I think the new people in the party are really enthused by Bernie Sanders. Sanders was able to capitalize on his surge, but I think Clinton has been able to beat that back. On the Republican side, it’s been all over the place. Read more