Ask most political party strategists, U.S. pollsters and experts and they’ll tell you that the 2024 presidential election could be the closest in decades, if not 100 years.
That’s a plausible argument, considering Joe Biden’s 2020 victory was decided by fewer than 45,000 votes in just three battleground states.
But that’s what the polls suggest. The race between Kamala Harris and Donald Trump is extremely close nationally and in key states.
So how reliable are opinion polls?
This is a question many have been raising after they famously made a huge mistake four years ago.
The American Institute for Public Opinion Research showed that Biden’s lead over Trump in the final two weeks of the campaign was twice as large as when votes were counted, marking the largest missed vote in 40 years.
In 2012, pollsters significantly underestimated Barack Obama’s lead over Mitt Romney. But while Trump’s 2016 victory over Hillary Clinton shocked many, the margin of error in national polls was relatively small.
Of course, all polls involve uncertainty and some margin of error. Additionally, they are only a snapshot of the current situation and do not predict election day results.
But importantly, there is also evidence that they become more reliable predictors of the outcome as the election approaches. It may seem obvious, but a year after the vote, polls differ by an average of seven points from each candidate's final tally. By the final week of the campaign, this percentage drops to less than 3 points.
Daily polls give you a little more insight into the results.
So how should we read it?
One important thing to remember is the uncertainty surrounding a candidate’s estimated approval rating.
Uncertainty: “margin of error”
Opinion polling organizations publish figures for each percentage, but they also report a “margin of error” that shows how much support rates fluctuate.
For example, if a poll shows Trump’s approval rating at 46% (with a margin of error of 3 percentage points), that means Trump’s approval rating among voters should be between 43% and 49%. If Harris’s approval rating is 49% in the same poll, her approval rating should be between 46% and 52%.
All this tells us is that the race is close and either candidate could lead.
Similar precautions should be taken when verifying trackers using polling averages.
You might think that averaging the surveys would reduce uncertainty, since random errors should cancel out. But while some polling agencies are consistently more accurate than others, some polling agencies may be systematically wrong in one direction. Adding all of these together can reinforce bias.
Read more about the election:
What will happen on the night of the US presidential election?
How does Donald Trump maintain such strong support?
Image: President Trump in Wisconsin at the beginning of the month. Photo: Associated Press
Predicting turnout: context matters
Who votes in an election also matters to the outcome, but it’s difficult for pollsters to predict. About one-third of eligible Americans do not vote in presidential elections, and the same people do not vote every time.
Context matters. That could make people more or less likely to vote. For example, potential abortion law reform appears to have mobilized many Democrats for the 2022 midterm elections.
A candidate’s policies and track record can also change the likelihood that more partisan voters will turn out.
The Electoral College: Why state races matter
Similar considerations are necessary when looking at state polls, and perhaps even more important in determining which candidate has the best chance of winning an election.
Presidential elections are decided by the Electoral College on a state-by-state basis, but the gap between them and the national vote is widening.
Read more: What is the Electoral College?
Image: Harris discussing abortion rights in Arizona in June. Photo: Reuters/Rebecca Noble
In 2000 and 2016, the candidate who received the most votes nationally lost the election because he did not win the combination of states that received the most votes in the Electoral College.
This is why polls in swing states get so much attention and are incorporated into the models of statisticians trying to predict outcomes.
Of course, there are the same uncertainties and potential flaws as national polls. And the bad news is that their recent performance hasn’t been good.
In 2016, they suggested that Clinton would win the key states needed for an easy victory in the Electoral College. In 2020, their performance was even worse.
Polling has proven particularly difficult in some states. In 2016 and 2020, the biggest failures were Wisconsin, Michigan, North Carolina, and Pennsylvania.
All of these, especially Pennsylvania, which has the most electoral votes of the group, could have a decisive impact in 2024.
Follow Sky News on WhatsApp
Follow Sky News for the latest news from the UK and around the world.
tap here
So what do the polls tell us?
If you want to know how people feel about a particular candidate or policy, there are plenty of answers.
But if you’re trying to determine who’s ahead in the presidential race, we can say with confidence that the race is very close, and the number of votes could be close in some states. That’s all.
