Something strange is happening in the world of polls.
In short, FiveThirtyEight's model gives Joe Biden a much better chance of beating Donald Trump this fall than other leading models, and its odds are about as good as Biden's improve Ever since his aimless performance at the June 27 presidential debate.
Take Wednesday as an example:
-
FiveThirtyEight's model puts Biden's chance of winning the election at 54 in 100. (That is, win enough states and he will win the Electoral College, no Win the popular vote.
-
By contrast, pollster Nate Silver founded FiveThirtyEight but left last year with his own model that gave Biden… say significantly reduced Chance to win. (To see the exact numbers, you must subscribe to Silver's Substack.)
-
The Economist's model shows Biden has a one in four chance of winning the election.
-
The New York Times polling average shows Trump leading by 3 points or more in six swing states that Biden won in 2020: Arizona, Georgia, Michigan, Nevada, Pennsylvania and Wisconsin. Much – if accurate, this gap would amount to a huge gap.
The gap between FiveThirtyEight and its rivals is important given how many Democrats believe Biden should end his re-election bid as polls show him facing significant losses.
Biden has responded by repeatedly claiming that the race is actually too close, a position that FiveThirtyEight, perhaps the most prominent election forecaster, appears to confirm. On Tuesday, former Biden chief of staff Ron Klain tweeted a graphic of the site's electoral vote projections, lending credence to claims from donors and elected Democrats that the president lacks a “path to victory.” Incorrect. Simon Rosenberg, a pollster who strongly supports Biden, said something similar.
On the other hand, no one agrees. Other pollsters have begun asking FiveThirtyEight's current director, G. Elliott Morris, direct questions about his methodology. (Slate reached out to FiveThirtyEight and its parent company, ABC, which declined to comment.) So what happened? I asked David Broockman, a professor of political science at the University of California, Berkeley, who teaches a quantitative methods course, what he thought.
“I don't have the code for their model, so I'm not sure, but my understanding is that they're basically putting all the weight on fundamentals now, so pretty much no matter how bad the polls are, they're going to show the same Answer,” Brookman wrote in response. In prediction, basic knowledge Refers to factors independent of polls — unemployment, inflation data, people in office — that have historically been linked to election outcomes. Biden is the current president overseeing a growing economy, which should mean he's well on his way to re-election.
But he's not. Other pollsters seem to be paying less attention to fundamentals, which means they are paying more attention to polling, and Biden is (without a doubt) losing the polls.
Perhaps Biden is more concerned, Brookman said, that there are good reasons to believe fundamentals cannot help predict the outcome of this race as they have historically. “The rationale is, historically, early polls have not been very predictive. But that's because people don't know the candidates yet,” he said. “It's hard to imagine people not knowing how they feel about these two!”
If the fundamentals are favorable for Biden and he still fails, this would seem to prove that a generic alternative candidate is expected to do just that. better one Worse than him, not that bad – especially considering his apparent limitations as a communicator. (In other words, previous candidates who enjoyed the same fundamental advantages as Biden may have been better equipped to defend themselves on television.) Beyond that, statistically-minded observers are also concerned that FiveThirtyEight’s state-by-state forecasts The quirks raise additional questions: suggesting the site overstated Biden's chances for reasons beyond the base weight choice.
Regardless, FiveThirtyEight's model becomes increasingly unfavorable for Biden as Election Day approaches, and it places less emphasis on fundamentals. (Other models will do this as well.)
Even the folks at FiveThirtyEight say the same thing. Morris wrote on Twitter/X that the site's version of the model places less emphasis on fundamentals and currently has a 57% chance of Trump, not Biden, winning. (The same tweet added that “none of this will matter that much by Labor Day” because of the predictive value of polls relative to other information.) He also wrote that Klain cited his model for Biden’s The rationale for staying in the race is, quote, “weak sauce.” At least at this point, everyone involved in the polls seems to think agree.
