Adventures in Polling, Chapter 437
December 2023
Why should I trust the polls? That question comes up at virtually every webinar or speech I’ve given over the past seven years.
After Donald Trump’s victory in 2016, Joe Biden’s closer-than-expected win in 2020, and “the red wave that wasn’t” in 2022, it’s a valid question. But, setting aside whether expectations correctly matched the polling, there has been a bit of an overreaction — even an overcorrection — to a point that it’s possible to miss the story that polling is trying to tell.
Trump is the clear front-runner in the GOP presidential primary. How do we know? Because the polls tell us he is.
Trump is leading in the national polls. He had 62% in FiveThirtyEight’s national average on Dec. 12, 50 points ahead of Florida Gov. Ron DeSantis (12%), former U.N. Ambassador Nikki Haley (11%) and the rest of the field. But, of course, we don’t have national elections in this country.
More important, Trump is also the clear front-runner in the early primary states, including Iowa. With 48% in the FiveThirtyEight average as of Dec. 16, he outpaced DeSantis (19%) and Haley (16%) by a wide margin. The Des Moines Register poll, which is considered the gold standard in Iowa, had Trump with 51% in an early December survey followed by DeSantis (19%) and Haley (16%). Trump led the field by 27 points in New Hampshire in the RealClearPolitics average as of Dec. 12.
Everyone seems to be so snakebit by the 2016 result that they’re too cautious about projecting races in 2024, even in the face of clear data.
Specifically, if there were going to be a surprise in Iowa, we would start seeing signs of that by now. Looking back at polling in mid- to late December 2015, Texas Sen. Ted Cruz led Trump 24% to 19% in a Monmouth University survey, 31% to 21% in a Des Moines Register/Bloomberg poll, and 30% to 23% in a Loras College poll. In the end, Cruz won 28% to 24%. Dare I say, polling was pretty good that far out from the election. And those 2016 caucuses were two weeks later (Feb. 1) than they are this cycle (Jan. 15).
Wins by former Pennsylvania Sen. Rick Santorum and former Arkansas Gov. Mike Huckabee were more of a surprise in 2012 and 2008, respectively, but those were wide-open races without a de facto incumbent like Trump, who is significantly ahead of the field this time and is a known commodity, and there were signs of their surge by this point as well.
More generally, projecting a Trump loss would require a 180-degree shift in how we look at polling. For the past seven years, we’ve been conditioned to assume that polls underestimate Trump’s support. But in order to believe that DeSantis or Haley or someone else is going to win this nomination, we’d have to believe that the polling is overestimating Trump’s support. It’s technically possible, but seems pretty unlikely.
Another recent round of polling is being dismissed prematurely.
In early November, Democrats had a great night reelecting the governor of Kentucky, capturing full control of the Virginia legislature, expanding their House majority in New Jersey, passing a key abortion access measure in Ohio, and even holding the Republican governor of Mississippi to a 3-point victory.
Some people believed those results were incongruent with New York Times/Siena polling that showed Biden losing to Trump nationally, and Trump leading Biden in Georgia, Nevada and Michigan and running even with the president in Arizona and Pennsylvania. But the election results and the polling are reconcilable.
Presidential polling was not meant to measure or predict the 2024 results. I don’t have any doubt that, if the presidential election were held today, Trump would win, particularly when armed with a recent Wall Street Journal survey that showed Trump ahead of Biden in a head-to-head matchup as well as having the confidence of voters on key issues compared with the president.
But the presidential election is not today. If Biden goes on to win the election in 2024, that doesn’t mean polling in November and December of 2023 was inaccurate. It means the circumstances of the race changed.
Polling still has its methodological challenges and should be digested with a dose of skepticism. But sometimes we don’t need to overthink it, and we can believe the data when it’s pointing in a certain direction.
Nathan L. Gonzales is a senior political analyst for the Public Affairs Council and editor of Inside Elections, a nonpartisan newsletter with a subscription package designed to boost PACs with a regular newsletter and exclusive conference call. You can also hear more on the Inside Elections Podcast. His email address is [email protected].
“
If Biden goes on to win the election in 2024, that doesn’t mean polling in November and December of 2023 was inaccurate. It means the circumstances of the race changed. Polling still has its methodological challenges and should be digested with a dose of skepticism. But sometimes we don’t need to overthink it, and we can believe the data when it’s pointing in a certain direction.
More News & Resources
Featured Event
Is your organization prepared to adapt its social impact initiatives based on who is elected, from the President all the way down-ballot? Navigate post-election shifts at STRIDE this November.
Washington, D.C. | November 21