By John A. Tures, Professor of Political Science, LaGrange College

There’s a myth that the polls were wildly inaccurate in 2024, a common refrain that we’ve heard since 2016. Are the polls so inaccurate? Are they getting better or worse? And how did they perform in 2024? And did polls play a role in the election outcome?

According to the RealClearPolitics site, Donald Trump got 50.47 percent of the vote in 2024, while Kamala Harris received 47.9 percent of the vote, a difference of about 2.5 percentage points. On that same site, poll average showed the race was almost dead even. That’s pretty much within the margin of error of most surveys, many of which are three percentage points or greater.

I looked individually at the polls since the Fox News poll from October 11-14 until just before Election Day, by Atlas International, TIPP and Ipsos, again collected by RealClearPolitics. Only one poll from NPR/PBS/Marist from October 31-November 2, was outside the margin of a 3.5 margin of error of 3.5 percent or more. That poll had Harris up by four points. Five others had Harris up, but only up by two points or less. Five had Trump up by three points or less. Five had the race tied.

In other words, 94 percent of polls were within the margin of error of 3.5 percentage points. It was a close election in the polls, and that’s what surveys showed. And the results in the popular vote were close too. My research methods students who looked at swing states, and “swing states adjacent” as I like to call them (Florida, Ohio, Minnesota, Texas, New Hampshire, Virginia) found that many of these state polls were also pretty close between the two nominees.

In a column for Raw Story, I covered the quality of the polls over time. In 1936, the Literary Digest predicted Alf Landon would defeat President Franklin Delano Roosevelt by nearly twenty percentage points, when FDR was able to win by a similar margin. I also noted huge errors in the past, like the 1948 election, with numbers back then close to 10 percentage point deviations.

But in that Raw Story column, I also discovered that in recent elections, the margins of error were 3.1 percentage points in 2020, 4 percentage points in 2016, 2.9 percentage points in 2012, and 1.6 percentage points in 2008. So the 2024 Election wasn’t just better than surveys from the 1930s and 1940s. They were in pretty good company for the last several election cycles. Of course, there’s always room for improvement, but claims that “the polls were way off” aren’t supported by the evidence.

But there is a way that the polls affected the election: the all-too-frequent reporting of them. It seemed that there were so many articles that covered nothing but polls and their results on a nearly daily basis, to report on what we already knew, which was the race was pretty close. What wasn’t covered was Kamala Harris’ proposed policies, or what “Agenda 47” of Donald Trump’s plans would do, and whether it was different from Project 2025.

“I wish there was more coverage of what the candidates would do for education,” a respondent told me while I was conducting exit polls. And that’s how polls affected the election: their frequent coverage drowned out a substantive discussion of the issues.

One way we could improve this is to copy what they do in the NCAA college sports. They release a weekly survey showing the team rankings, but they only do it once a week, instead of once a day. That would leave the other six days to cover election issues.

John A. Tures is a professor of political science at LaGrange College in LaGrange, Georgia. His views are his own. He can be reached at jtures@lagrange.edu. His “X” account is JohnTures2.