National polls for the US presidential election have changed significantly since Kamala Harris took over from Joe Biden to face off against Donald Trump.
While Biden has lagged the former Republican president nationally and in many key battleground states, Harris has risen about three points in national polls since becoming the nominee.
The Guardian’s poll tracker, which assesses 10 days of rolling polls, currently shows Ms Harris leading nationally by about two points.
But comparing the results to past elections, data from RealClearPolitics shows that Harris’ lead over Trump is weaker than her past opponents: Hillary Clinton led Trump by 5 points in 2016 and Biden by 6.3 points in 2020, as of August 30.
But national polls only tell part of the story: Clinton lost the 2016 election because of the system that selects the president, even though she received more votes than Trump.
Several battleground states are leaning to Harris, making the race close.
The US presidential election is decided by the election of each state that has a certain number of electoral votes, which means the election is decided by a small number of battleground states such as Pennsylvania, Arizona, and Georgia.
According to data from RealClearPolitics, Harris has at least managed to close the gap in battleground states, even if she hasn’t managed to overtake Trump, especially in Georgia and Arizona, where she has increased her vote lead by more than 4 percentage points since Biden dropped out of the race.
Pollsters have argued that these states have become hotly contested again because Harris has solidified support among the Democratic base, particularly voters of color and young people who left the party in the final months of Biden’s presidency.
A few points ahead of Trump nationally isn’t enough to give Harris a solid lead in any of the battleground states.
Trump and Harris are still within two points of each other in seven battleground states, within the poll’s margin of error.
That’s key, said Christopher Bolick, a political science professor and director of the Muhlenberg College Institute for Public Opinion Research. “No matter how this election goes, certain things are going to matter.”
“The 2016 and 2020 polls weren’t that far off, but the points they were off were big. The margin of error doesn’t determine the winning line, and small errors can make a big difference.”
This article analyzes the latest polling averages from RealClearPolitics, but other polling averages from Silver Bulletin and FiveThirtyEight show Harris with slight leads in Michigan, Georgia, Nevada, Pennsylvania and Wisconsin.
Past polling errors cause concern for Democrats
While Harris has enjoyed favorable polling in recent weeks, there’s another uncertainty to consider: Trump polled well in 2016 and 2020.
There are several reasons for this: many people made their decisions very late in the campaign, many Trump supporters did not tell pollsters who they supported, and methodological errors led to a fundamental underestimate of support for Trump.
Josh Clinton, chair of the American Association for Public Opinion Research (AAPOR) task force that developed the 2020 polling study, told the Pew Research Center: “The polling error was the largest in 40 years for the national popular vote and the largest in at least 20 years for state-level vote estimates.”
In 2020, in the case of FiveThirtyEight in Wisconsin, certain polling averages underestimated the outcome by as much as 7.7 points. While the pollster had Biden leading by 1.8 points in North Carolina, RealClearPolitics had Trump ahead by just 0.2 points. Trump actually won the state by a 1.3 point lead.
Methodological errors included imprecise weighting that led to under-representation of less educated voters and a failure to take into account voters’ remembered voting patterns. Pollers have sought to address these issues in this election, for example by broadly adopting education level as a survey weight.
In addition to these adjustments, 2022 gives people hope: According to an analysis by FiveThirtyEight, 2022 was one of the most accurate polling years in recent history.
But Trump’s return to the race has increased uncertainty.
“Trump’s support may have been systematically underestimated. We’ve seen this in our polling, and I think it’s reasonable to consider the possibility of it happening again in 2024. Adjustments we made in 2020, such as weighting education levels, ultimately helped, but they didn’t fully correct for the dynamics that we believe Trump is creating and don’t fully understand,” Bolick said.
“When Trump runs, the difference is clear. While you should be skeptical of a sample size of two presidential elections, there seems to be a gravitational pull to a Trump election.”
How the Race Will Turn Out
As it stands, website 270toWin classifies seven states as “close,” projecting a close election with 226 Democrats and 219 Republicans winning.
If those states vote as expected, Harris would only need to win three of Biden’s battleground states — Michigan, Pennsylvania and Pennsylvania — to secure the 270 electoral votes to win, while Trump would need to win at least two states from the Democrats to win.
That leaves uncertainty over the close outcome of the 93 electoral votes and the possibility that polls may still be underestimating support for Trump.
So while Harris may be pleased with the momentum her campaign has enjoyed over the past month, the outcome of the election remains to be seen.