Even the Best Pollster Missed Iowa by About 13 to 15 Points
A celebrated Iowa poll showed Harris ahead, then missed by double digits.

The Ezra Klein Show · Nate Silver Predicts a ‘Momentous’ MidtermOn the Saturday night before the 2024 presidential vote, election forecaster Nate Silver had dinner plans when a friend texted with a rumor about Iowa.
The rumor was that Ann Selzer, the longtime Iowa pollster known for nailing Obama's strong showing in the 2008 caucuses, was about to release a survey putting Kamala Harris three points up in Iowa. Other surveys in the state showed Donald Trump ahead by about nine, and Silver's statistical model also pointed the other way.
Silver used the episode to argue that polls are judgment calls that get revised, not detached measurements. Pollsters weight for age and party, fill gaps left by unanswered calls and online samples, and guess who will actually vote.
Selzer stood at the top of Silver's ratings, tied with The New York Times for first place, which made the conflicting signals hard to ignore. Her final Iowa number missed by what Silver guessed was 13, 14 or 15 points, which is why the size is given as a range.
Unwilling to bet outright against her, Silver nonetheless stuck with his data rather than reshaping it to fit her result. He wondered if rivals had overcorrected for missing Trump before, but the Iowa finding did not square with everything else.
Ezra Klein, the journalist interviewing Silver, called Trump's response insane. Trump sued the Iowa newspaper that published the survey, treating an unfavorable result as a basis for court action.
Silver put the larger point this way: “a poll is kind of almost like a mini model with polling-flavored inputs.” A forecaster by his own description, not a pollster, he meant surveys blend data with choices about sampling and weighting.
Klein admitted those adjustments leave him disoriented, from demographic weighting to turnout estimates. Silver answered that the line between polling and forecasting keeps blurring.
That blending is why methods keep shifting, he argued. Like a restaurant that tones down a spicy dish after complaints, then hears regulars miss the heat, pollsters adjust after misses because reputations ride on the final weeks. Yet sudden shifts can still blindside them; in 1994, he noted, what seemed a tough year for Bill Clinton's Democrats became a far worse one.
Questions people ask
Who was Ann Selzer?
Selzer ran a respected Iowa polling operation for decades and became famous for accurate calls in hard to poll races, including Obama's 2008 caucus showing. Silver said he believed she ranked at the very top of his list. She stepped away from election polling after the 2024 miss.
How can a professional poll miss by that much?
A poll interviews a small sample and then weights the answers to match the expected electorate by age, education, party, and other traits. If the people who respond are different from the people who vote, or if the turnout guess is wrong, the adjustment can fail. A small, unrepresentative sample can then produce a large error.
What is the difference between a pollster and a forecaster?
A pollster conducts the surveys and makes choices about sampling and weighting. A forecaster builds statistical models from many polls, economic data, and past results to estimate likely outcomes. Silver said he is a forecaster, while noting that the distinction is becoming blurrier.


