So what’s up with that Iowa poll? A few scenarios.

[cigaretteman]

For the past decade, the Des Moines Register’s final poll of Iowa before a presidential contest has mostly drawn attention when it comes in January or February — that is, before the Iowa caucuses set presidential contenders on the path to the nomination (or, often, not). The polls released before the 2016 and 2020 general elections were less momentous because Donald Trump had held consistent, stable leads in the state. And those held.


Cut through the 2024 election noise. Get The Campaign Moment newsletter.

But the results of the Register’s final 2024 poll, conducted by Selzer & Co. and released Saturday evening, were truly stunning, even when considering the high bar required for any new development to reach that standard in this tumultuous year. It wasn’t just that Iowa was close, which would have been remarkable. The poll showed Vice President Kamala Harris with a narrow lead, albeit one that still sits within the margin of error.
Skip to end of carousel

Sign up for the How to Read This Chart newsletter​

Subscribe to How to Read This Chart, a weekly dive into the data behind the news. Each Saturday, national columnist Philip Bump makes and breaks down charts explaining the latest in economics, pop culture, politics and more.

End of carousel
Betting markets reacted immediately. Social media lit up. But everyone was wary: Was Harris really going to win Iowa? Was the result an outlier, a poll result that is simply an anomaly? Or was something else happening?


ADVERTISING


So let’s consider three scenarios for what the Iowa results might mean.

1. Harris is in position to win Iowa.​

We will start with the most obvious possibility: The poll, from Iowa’s most respected pollster and one of the most trusted pollsters in the country, is right. Harris led 47 percent to Trump’s 44 percent in the survey and, while Selzer & Co.’s Ann Selzer acknowledged in an interview with Reuters that late shifts could occur, under this scenario a slight Harris victory would be possible.
🏛️
Follow Politics
One reason this idea seems at all feasible is Selzer's reputation. Here are the final Register polls in recent presidential elections and the actual outcomes.

Year	Poll	Result
2004	D+5	R+1
2008	D+17	D+10
2012	D+5	D+7
2016	R+7	R+9
2020	R+7	R+9

There are two things in particular to notice there. The first is how accurate the polling since 2012 has been. The second is how recently a Democrat, Barack Obama, won the state.


Understanding that the result of the 2024 poll would be unexpected, Selzer and the Register have noted that the current margin follows a trend seen in previous polls of the state. In June, Trump was leading President Joe Biden by 18 points. By September, that lead (now over Harris, after Biden dropped out) had shrunk to four points, a 14-point shift. The latest margin continues that trend, shifting seven more points to the Democrat.

https://www.washingtonpost.com/politics/2024/11/04/so-whats-up-with-that-iowa-poll-few-scenarios/

 

[cigaretteman]

2. The poll is an outlier.​

There’s a famous article from the satirical newspaper the Onion in which two people are debating the effects of invading Iraq in 2003. One man’s position is, “This War Will Destabilize The Entire Mideast Region And Set Off A Global Shockwave Of Anti-Americanism.” His opponent’s position is, “No It Won’t.”

To some extent, waving away the Selzer results as an outlier is simply No-It-Won’t-ing the result. It’s hard to tell in the moment if a poll with an unexpected result is an outlier or is picking up on a trend that other pollsters have missed or haven’t yet registered.




As the New York Times’s Nate Cohn wrote on social media, “Selzer can be wrong, and has been before.” He added that “in the end everyone in this business is subject to sampling error and so on.”
“This Selzer Iowa poll is off on its own,” he wrote in a subsequent post, “not just in Iowa but in terms of the overall story.”
He was writing that in October 2020, not now, about a Selzer poll that nailed the result in Iowa.

3. The poll captures something missed elsewhere.​

The national story being told in 2020, Cohn argued, was that Biden was widely outperforming how Hillary Clinton had done with White voters in 2016. In Iowa, that’s most voters.

What’s happening now in Iowa, Selzer explained to the Register, is that “age and gender are the two most dynamic factors that are explaining these numbers.” Older women in particular overwhelmingly favor Harris, according to the poll, by a 2-to-1 margin.


The Register calls this a “late shift,” though it’s not clear if they mean late in the sense of “since our September poll” or in the sense of “as the election period wraps up.” There have been suggestions that Harris is benefiting from people who have only recently made up their minds; a New York Times-Siena College poll of swing states released over the weekend suggested that Harris had an 11-point lead among those late deciders. Late shifts can upend poll-based predictions, as they did to an extent in 2016. Polling is necessarily old by the time it becomes public, and the shift to Trump as Election Day arrived in 2016, particularly among independents, wasn’t captured in surveys.


https://www.washingtonpost.com/poli...d=mc_magnet-election2024_inline_collection_15

What may be more important is the extent to which women are driving the results in Selzer’s poll.

We should back up for a moment and remind you how polls work. Pollsters contact various people, hundreds of them across the area being polled (in this case, Iowa). But it’s almost impossible to get a perfectly representative sample of responses, a pool of respondents that looks exactly like the target population in gender, age, race and education. So pollsters weight their results, giving respondents’ answers more or less importance in the results, depending on how representative those respondents are.


Importantly, the target population here isn’t Iowa — it’s Iowans who will vote. So the pollsters need to weight the results, but to do so to who they think will actually cast a ballot. If their assumptions about the electorate are off, their results could be, too. (Selzer explained how this poll was weighted in an interview with CNN on Sunday night.)
Last month, I noted that the shift to the right among non-White voters seen in the Times-Siena polling over the course of the year — a shift that triggered a flood of attention — was counterweighted by a shift in favor of the Democrats among White women, a much larger part of the electorate. If women (and White women in particular) are being underestimated as a part of the electorate, Harris’s position might seem weaker than it actually is.

This has been an argument made since the decision in Dobbs v. Jackson Women’s Health Organization that overturned Roe v. Wade. Special elections, particularly those focused on access to abortion, saw unexpected turnout and positive results for Democrats and pro-access positions. The 2022 midterms were better than expected for Democrats, with credit being given to those voters. It has consistently been possible that the 2024 polls have similarly underestimated enthusiasm among women — perhaps in part because pollsters have been wary of once again being seen as underestimating Trump.
Perhaps Selzer’s result in Iowa will prove to be overly favorable to Harris. But maybe the poll is also getting closer to capturing a shift that other polls haven’t incorporated.
There are a lot of unknowns here, including which of the three scenarios above is most accurate. Happily, most of those unknowns will be known in about 48 hours.