Among the many moving parts of elections, the behavior of undecided voters can be among the toughest to suss out. Some don’t make up their minds until the last few days of a campaign, even on the last day. Some never do and either don’t vote at all or skip over voting for that office.

Many political observers prefer to look at polls in a binary way: They are either right or wrong. To be sure, even the best of pollsters are off from time to time, but statistically speaking, only about 5 percent of the time are poll results outside the margin of error from what they would be if every voter were questioned. These binary thinkers don’t want to consider the undecided voters, those who had not made up their minds when they were interviewed for a poll.

But pollsters are not mind readers, and polls are not crystal balls. As the industry cliché says, polls are “a snapshot in time,” but they do not see over hills or around corners. If undecided voters break disproportionately for one candidate over the other, that does not make all polling taken before those voters made up their minds “wrong.”

Theoretically, undecided voters can choose not to vote; they can vote for a third party, an independent candidate, or a write-in. As a group, they can split more or less down the middle, or they can break either left toward Democrats or right toward Republicans.

Exit polls conducted in the presidential election four years ago showed that those who made their minds up relatively early in the campaign tended to vote for Hillary Clinton, while those who made their decisions late cast their ballots more for Donald Trump. Some people theorize that the late break was related to FBI Director James Comey’s letter or Russian interference in the election, but whatever the case, how are pollsters supposed to be able to determine that?

Recall the election of 1980, which was considered too close to call in its late stage. A large pool of undecided voters were not inclined to reelect President Carter but were also hesitant to vote for Ronald Reagan, not sure if the former movie actor was up to the job (notwithstanding his two terms as governor of our largest state). But after the only debate featuring both Carter and Reagan, exactly a week before the election, undecideds took in Reagan’s strong performance and shifted in the former California governor’s direction, leading to a 51-41 percent win (John Anderson pulled 7 percent). Again, were polls taken earlier wrong?

Prior to the elections this year, polls showed Joe Biden dancing around the 50 percent threshold. The final RealClearPolitics poll average had Biden ahead 51 percent to 44 percent. Fox News had the race at 52-44, while NBC News/Washington Post put it at 52-42. Now, as of late Thursday afternoon, the actual national popular-vote count gives Biden 50.8 percent to Trump’s 47.4 percent, a 3.4-point margin. It would seem, then, that much of the polling came close to nailing Biden’s share of the vote but in the end the undecided voters broke for Trump, as they did four years ago.

This column speculated last week that a slice of voters who seemed intent on voting not just against Trump but also to punish his party may have had second thoughts about putting Democrats in charge of everything.

This year, the undecideds broke moderately back toward Trump, making the race closer than it seemed to be earlier, but not enough to save him from losing five states he won in 2016—Arizona, Georgia, Michigan, Pennsylvania, and Wisconsin—as well as Nebraska’s 2nd Congressional District. In congressional elections, the undecideds clearly broke toward Republicans, with some who cast ballots for Biden coming back over to vote for GOP candidates for the House and Senate.

What could have triggered the bout of cold feet among this narrow but apparently pivotal slice of voters? Could it have been all this talk of socialism, defunding police, Medicare-for-all, racial tension, or episodes of urban violence over the summer? Research by Democratic pollster Stan Greenberg points in that direction. Toss in, perhaps, some fear that a Democratic president and big Democratic majorities would, if they couldn’t win the game, simply change the rules by packing the Supreme Court or ending the Senate filibuster. In any case, the drop-off in votes from Biden to votes for Democratic House candidates was about 4.2 million nationwide; for Republicans, it was only about 1.8 million.

Political strategists from both parties, pollsters, focus groups, and campaign operatives in the field were seeing in the closing weeks of the campaign the same things the rest of us were seeing, particularly in the suburbs. No less an authority than Sen. Ted Cruz, in a CNBC interview a month before the election, pointed to an incredibly volatile political environment, warning that the election “could be a bloodbath of Watergate proportions” for his party. So just because the wave didn’t make it all the way to the beach didn’t mean it wasn’t out there all along.

This article was originally published for the National Journal on November 13, 2020.

More from the Cook Political Report

CPR 2024 National Polling Average
Cook Politcal Logo