When elections don’t go exactly as expected, everyone is quick to point fingers at pollsters for getting it wrong. And no one takes more heat than FiveThirtyEight founder Nate Silver.
During a punchy episode of FiveThirtyEight’s now daily politics podcast on Wednesday, Silver pushed back forcefully on anyone out there accusing him or his website of getting the 2020 election wrong by predicting that Joe Biden had a 90 percent chance to win what has ended up being a nail-biter of a race.
“I do want to ask about polling, because I think I would be virtually decapitated on Twitter if I didn’t get into this,” podcast moderator Galen Druke said about 20 minutes into the group’s discussion, raising the significant polling errors in states like Florida and Wisconsin. “The pitchforks are already coming for the pollsters and the prognosticators,” Druke said, asking his boss to address the “rage” that has been directed his way over the past 24 hours.
ADVERTISEMENT
“If they’re coming after FiveThirtyEight, then the answer is fuck you, we did a good job!” Silver replied, explaining that the only reason former Vice President Joe Biden was considered such a heavy favorite in the site’s closely watched model was because “he could withstand a 2016-style polling error or a bit larger” and still win the election. On the eve of that election, FiveThirtyEight gave Donald Trump a 28.6 percent chance of winning, much higher than other polling aggregators.
Silver also pointed out that the only state where Biden led in the final polling averages and lost at this point was Florida, with North Carolina likely to follow. And Silver said he found it “a bit concerning” that there have now been significant polling misses “in the same direction” not only in 2016 and 2020 but also in the 2014 midterms (though not in 2018).
Appearing on ABC’s This Week with George Stephanopoulos on the Sunday before Election Day, Silver made similar arguments, noting that Biden would still likely win the race even if there was a polling error as big as 2016. Meanwhile, he added, if Trump did end up winning, Silver said he thought it would come down to the tipping point state of Pennsylvania, which still hangs in the balance as of Wednesday night.
“People have false impressions of how accurate polls are, and the fact that Biden had this big lead was why he was a pretty big favorite, not because polls are perfect,” he concluded on the podcast.
In response to Silver’s defiant defense of FiveThirtyEight’s model, Druke promised that the site and podcast would be taking a much closer look at why some polls were so wrong this time. “But we’re not going to do it late at night after we haven’t had any sleep,” he said.
