• D.C.
  • BXL
  • Lagos
  • Riyadh
  • Beijing
  • SG
  • D.C.
  • BXL
  • Lagos
Semafor Logo
  • Riyadh
  • Beijing
  • SG


icon

Semafor Signals

Analysts debate late, contradictory polls in US election

Nov 3, 2024, 1:38pm EST
politicsNorth America
Democratic presidential nominee Vice President Kamala Harris.
Jacquelyn Martin/Pool via Reuters
PostEmailWhatsapp
Title icon

The News

With the 2024 US presidential election finally at hand, the final weekend of the race has seen a slate of last-minute polls that seem to offer more questions than answers.

Driving the discourse was a Des Moines Register poll that found Vice President Kamala Harris is leading former president Donald Trump by a 47-44 margin in Iowa — a shocking result for a state that had been considered a lock for Trump. The pollster, J. Ann Selzer, is considered the “gold standard,” having accurately predicted the outcomes of several past elections in Iowa.

AD

Other polls tell different stories, but with 48 hours or so to go until Nov. 5, virtually every prominent pundit seems to agree: For all the polls, no one can say with confidence who is going to win.

icon

SIGNALS

Semafor Signals: Global insights on today's biggest stories.

If accurate, Selzer poll signals massive midwestern shift

Source icon
Sources:  
Semafor, The Washington Post

Selzer’s Iowa poll could signal that Harris is outperforming among white midwestern voters. That could have major implications for the nearby swing states of Wisconsin and Michigan, which have similar demographics. Trump won Iowa by 8 points in 2020, and if he loses the state this time around, so the theory goes, then his chances of winning the presidency evaporate. Women are driving Harris’ lead, especially older or independent women, a trend many analysts say is driven by concerns over abortion. But “it’s important to remember this is just one single poll,” and it “does not suddenly wipe away hundreds of others,” Semafor’s Benjy Sarlin wrote. In response, the Trump campaign pointed to another Iowa poll Saturday that showed the Republican up by 9.

Poll ‘herding’ may be to blame for all the confusion

Source icon
Sources:  
Nate Silver, Kristen Soltis Anderson

Selzer’s poll bucks prevailing wisdom in a race that has been defined by razor-thin swing state polls. That may be in part a symptom of poll “herding,” whereby polling firms trend toward a consensus, and just don’t publish polls that look like outliers. Herding can be a way of saving face, analyst Nate Silver wrote, enabling polling companies to avoid sticking their neck out with a surprising finding that turns out to be wrong. Selzer, by contrast, doesn’t herd her poll, and does little to weigh her model in one direction or another. If she’s right, “this is polling equivalent [of] the Battlestar Galactica being the only ship in the fleet to not use the super new fancy technology,” and thus the only ship to survive the Cylons’ initial attack, pollster Kristen Soltis Anderson wrote on X.

At the end of the day, no one knows what’s going to happen

Source icon
Sources:  
New York Times/Siena College, Nate Cohn, Amy Walter

Even as the Selzer poll suggested Harris made gains among white voters in the Rust Belt states, a New York Times/Siena poll Sunday showed contradicting results. The Times poll showed Harris making up ground with Black and Hispanic voters, while new Morning Consult polls were seen as good but not great for Trump. Ultimately, pollsters are stumped, with two days to go. “I have no idea whether our polls (or any polls) [are] ‘right’, too good for Harris, or too good for Trump. No one does,” The New York Times’ Nate Cohn wrote. Instead, Cook Political Report’s Amy Walter said, the final polls “give us a ’choose your own adventure’ ending to an unprecedented election.”

AD