Conventional wisdom among people who follow politics is that polling — especially at the presidential level — is getting worse.
That the decline of landlines, peoples’ refusal to answer the phone and the weird ways in which the electorate is both expanding and shrinking make polling the equivalent of throwing darts at a dartboard while blind-folded.
Not true, according to
, who writes the Substack. (You likely know Elliott from his time running 538.)Morris told me — in a Substack Live conversation Wednesday afternoon — that swing state polling averages in 2024 were actually far better than in 2020 or 2016
“Polls in 2024 were off by two points on margin in the swing states,” he said. “So on average, they said Kamala Harris would be tied or lead by one, and she lost by two or one, depending on the swing state…On margin in 2016, polls were off by four and a half and they were off by five in 2020.”
Remarkable right? I bet you didn’t know (or remember) that!
Morris has a simple explanation: Expectations.
In 2016, the expectation was that Hillary Clinton would win. She lost. So people concluded the polls were WAY off.
But, in 2020, the expectation was that Joe Biden would win. And he did. So even though the polls in swing states missed by MORE in 2020 than in 2016, people didn’t run around and scream about how polls were useless.
Last November, the conventional wisdom was that Harris would win — narrowly. Then Trump swept all 7 swing states and the presidency. And people, again, used that as proof the polls were all wrong. But the polls were more right than in either 2016 or 2020. What was off was peoples’ expectations.
“You get this weather prediction bias,” explained Morris. “Oh, the poll said...Kamala Harris is going to win, let's say. So other predictions said it was going to rain at 30% probability. And it didn't rain. And Donald Trump won the presidency, so it's useless. But that's just a difference in expectations.”
Elliott and I talked about a whole lot of polling-related stuff including:
Donald Trump’s polling floor and ceiling
Why it’s harder to poll Donald Trump than past presidents
His journey from The Economist to 538 to independent journalism
Why there’s no going back to the way journalism used to be
Our full conversation is FREE for all subscribers. Not one? Become one today!
And make sure to download the Substack app so you can watch future Substack Live conversations — I will be doing one or two every week — live!
Share this post