Trump supporters in Kenosha, Wisconsin, in 2020.
Photo: Scott Olson/Getty Images
While staring at the very close 2024 national and battleground-state polls, experienced observers are aware that polling was not exactly accurate in the last two presidential elections. But they were inaccurate in different ways: They pretty closely estimated Hillary Clinton’s national popular-vote lead over Trump in 2016 but missed Trump’s last-minute surge in the Rust Belt that gave him Michigan, Pennsylvania, and Wisconsin and the presidency. He won all three states by less than a single percentage point in what amounted to an electoral inside straight. In 2020, the polls significantly missed the national popular vote and battleground-state votes, though Joe Biden did win as expected. There are some mitigating factors, particularly the COVID-19 pandemic that affected both polling response and voting habits in 2020. And without question, pollsters have been steadily working on ways to reduce polling errors, especially with respect to mistrustful Trump voters in remote areas of the country.
But if you are inclined to be skeptical about the polls (or if you are a fearful twice-burned Democrat), it’s a good idea to drill down a bit and look at exactly where the most egregious polling errors occurred in 2016 and 2020, and where it might recur in 2024. Larry Sabato’s Crystal Ball took a close look at battleground-state polling and results in the two previous Trump elections, and it was pretty clear that in both these contests Wisconsin was ground zero for polling error.
Graphic: The Center for Politics
That’s right: The combined RealClearPolitics/FiveThirtyEight averages missed the actual results in Wisconsin by 7 percent in both 2016 and 2020; in both cases, the polling error was in favor of Democrats, which is how the polls missed Trump’s narrow win in 2016 and Biden’s narrow win in 2020 in that state. Michigan ranks second in polling error in both elections with a 4 percent overestimation of the Democratic percentage of the vote in 2016 and a 2.8 percent overestimation in 2020. Again, the polls missed a Trump win in 2016 and a closer-than-expected Biden win in 2020.
This is potentially unhappy data for Democrats since Wisconsin and Michigan are currently Kamala Harris’s best battleground states in the FiveThirtyEight averages: She leads in the former by 2.9 percent and in the latter by 1.7 percent.
On the other hand, Democrats came close to or exceeded their polling status in Georgia and Nevada: In the latter, Clinton outperformed her polling average by 2.4 percent in 2016 (perhaps a testament to that state’s legendary Democratic get-out-the-vote infrastructure); in the former, Biden beat his polling average by 0.1 percent, which is impressive since the national popular-vote polling error was 3.3 percent (averaging FiveThirtyEight and RCP) in Biden’s favor.
To be clear, the only thing the 2024 election has in common with 2016 and 2020 is the identity of the Republican nominee; Hillary Clinton and Kamala Harris have their own particular strengths and weaknesses, and pollsters may do a better job of finding and tabulating Trump voters this time around. There are also finely graded trends within all the battleground states that could affect the final numbers. But presently, if you are a Democrat, you’d like to see a bigger Harris lead in Michigan and Wisconsin, and if you are a Republican, Trump’s numbers in Georgia and Nevada could be suspect. It’s something to keep in mind as Election Day approaches.
Source link
