Nate Silver is a statistical expert and a professional poker player. His political prognostications are less than stellar, however, and should be viewed skeptically.
Silver gives Donald Trump a 55% chance of winning the election. He claims that's his "gut feeling" and not based on any statistical evidence. And he says that there's no statistical evidence that Trump has picked up any momentum and any polling results that say so are "questionable."
"Pollsters were 'herding' their numbers, or using past results to affect current ones, to keep Vice President Harris and former President Trump to within a point or two of each other each time" reports the New York Post.
“I kind of trust pollsters less,” Silver said on his podcast, name-checking Emerson College. “They all, every time a pollster [says] ‘Oh, every state is just plus-one, every single state’s a tie,’ No! You’re f***king herding! You’re cheating! You’re cheating!” He fumed.
Silver makes an interesting point. The 800 likely voter sample size of respondents is more than adequate. So why are the numbers in most of the seven battleground states within one or two points?
“Your numbers aren’t all going to come out at exactly one-point leads when you’re sampling 800 people over dozens of surveys,” Silver vented.
“You are lying! You’re putting your f***king finger on the scale!'”
Trump’s momentum in the polls has narrowed or erased Harris’s lead in some key swing states, but Silver says he thinks that after two cycles in a row in which the polls were “pretty bad” and plenty of criticism pointed at the polling industry ever since 2016, “a lot of pollsters are throwing up their hands.”
“They’re printing a tie,” Silver said at the CNBC event, giving the example of a hypothetical pollster conversation: ”’Harris is up 3%? Let’s say 1%. Trump up by 4? Let’s say 1.5%. There is what we call herding, where people are gravitating towards the consensus. They don’t want opinions too out of line.”
Another factor that could be skewing the polls is if Trump’s voters are more “charged up and excited,” which might lead them to partake in more polls than they have historically, and lead Harris-leaning voters to partake in polls less.
"Charged up and excited" Trump voters more willing to talk to pollsters? I call BS on that one. Trump voters show how "charged up and excited" they are by going to the polls on election day and voting. They hang up on pollsters, usually after a few well-chosen, unprintable epithets are shouted into the phone.
The last two election cycles have not fixed the polling problems of 2016 or even 2018 where many pollsters bragged about how good a job they did. Midterm elections are always more accurate because less than half as many people cast a ballot. They're more predictable by definition.
But on the day of the 2020 election, Joe Biden was supposedly winning by seven points. He ended up getting about half that margin more than Trump. Pollsters have tried earnestly to improve. It doesn't look like they'll do much better in 2024.
The polls in recent elections have been off by three to four points “systematically, across the board,” Silver said, so recent small swings in polls favoring Trump are “swamped by the uncertainty ... tripping off the noise, basically,” he said.
Five percent of people answer polls, “and they are weird,” Silver said.
Most people no longer even have the landline phones which past election cycle polling history relied on. “Old white people answer the phone, but others vote,” he added.
There's little doubt that Trump has been surging in the last two weeks. There's also no doubt that he was behind before his surge, especially in Pennsylvania, Michigan, and Wisconsin. It's probably still close in all three of those key states.
Silver is very brave for going out on a very long limb by declaring a pretty healthy margin on a bet that Trump will win. I don't think he or anyone else can say for certain, but I would be fairly confident in saying it's much closer than many in the GOP believe.
Join the conversation as a VIP Member