Four years ago at this time, polling pointed to a relatively easy victory for Joe Biden. On this day in 2020, for example, he led Donald Trump by 7.9 points nationally in the RealClearPolitics polling average and held advantages in all of the swing states, where some surveys even showed double-digit advantages for Biden.
Biden, of course, did go on to win the election, but by a smaller popular vote margin (4.5 points) and an exceedingly narrow one in the Electoral College, where a total of about 43,000 votes across three states separated him from a 269-269 tie that likely would have led to a Trump victory in the House of Representatives.
The race appears far tighter this time, with Kamala Harris just 2 points ahead of Trump in our current national polling average, and with the results in all seven battleground states within 2 points. It could be that the race is just as close as these numbers indicate and that polling this year is more accurately capturing the state of play.
But the possibility of another polling miss hangs in the air. If there’s a repeat of 2020 (and 2016) and the full extent of Trump’s support is again being missed, then he’s headed for a decisive victory. And if it’s Democratic support that this time is somehow being underestimated, then it’s Harris who’s actually in great shape right now.
Why Trump’s support was underestimated in the past is a matter of debate. But where it was missed is clear: states with large populations of white voters without four-year college degrees:
Wisconsin, where the polls were off by the most, has the highest concentration of white residents without college degrees of any battleground state. Michigan has the second highest. This was a key reason national polls were off, too.
Is it happening again? Here’s how the education divide among white voters looks in an average of current polls compared with what happened in 2020:
As you can see, Trump’s margin with noncollege-educated white voters ended up being higher than the polls suggested. But his current 27-point advantage is almost equal with that 2020 result.
Read one way, this buttresses the idea that polling is now largely capturing Trump’s support among white voters without college degrees. Then again, given how tight the polling is, it wouldn’t have to be off by nearly as much this time to make a big difference.
From Harris’ standpoint, part of the hope now is that polling is undercounting her support with what have long been core Democratic constituencies: Black, Hispanic and young voters. Here’s how these groups are breaking now compared with what played out in 2020:
The concern for Harris, obviously, is that her Hispanic support is far lower than Biden’s was, both in the 2020 polls and the final election results. But much of Trump’s new Hispanic support comes from younger voters who have not participated at high levels in past elections. If these voters end up sitting on the sidelines in this election, Harris could end up faring much better with Hispanics than the polling now shows. It’s also somewhat encouraging for her that Biden performed better in the election with Black voters than polling had suggested. Harris will need this to happen again.
Harris is also hoping that, in the post-Dobbs environment, her support among women is deeper than polling now finds. While current polls show her with an 11-point advantage with women, Biden had an 18-point edge in the 2020 polls and finished with a 15-point lead in the actual election.
Polling by gender has bounced around dramatically in this campaign, with some surveys showing historically wide gender gaps and others relatively modest ones. It averages out to an 11-point Harris lead among women, but plenty of polls show a much bigger advantage for her. If those polls are right, it could mean a Harris showing on Nov. 5 that is much stronger than the polls are now suggesting.
This story first appeared on NBCNews.com. More from NBC News: