What went wrong with the polls this year?

Tuesday’s election results delivered a second black eye for the nation’s pollsters in as many presidential contests as well as the unmistakable message that they have misjudged their ability to measure political opinion in an era of cellphones, polarization and Donald Trump.

With the vote-count still in progress, the size of this year’s polling error is still unknown. But as with the 2016 election, both national and state surveys left the impression that Election Night would bring clear Democratic gains, not a cliff-hanger.

No matter the outcome in the Electoral College, Democratic presidential nominee Joe Biden is likely to hold his lead in the popular vote, which stood at 2 percentage points as of Wednesday afternoon. Many analysts expect that lead to grow as Democratic-leaning states in the West complete their tallies. Unknown, however, is whether it will grow large enough to match the lead Mr. Biden held in the averaged, final results of national polls compiled by the websites RealClearPolitics or Fivethirtyeight.com, which on Election Day stood at 7.2 points and 8.4 points, respectively. The final Wall Street Journal/NBC News poll reported a Biden lead of 10 points.

While the final result may fall within the margin of error of national polls, opinion researchers said the outcome pointed to the need for evaluating their methods.

“We’re all trying to figure out where we go from here,’’ said Mark Blumenthal, a veteran pollster who consults with opinion research firms.

Polls of many House and Senate races—and the election analysts who rely on them—also appeared to be off target, as predicted Democratic gains in both chambers of Congress failed to come to pass.

Some pollsters had initial impressions about the cause of the missteps but said they needed to know more about the final results before digging in.

One theory gaining new attention was that renewed efforts this year to draw in more rural voters and those without four-year college degrees—groups supportive of Mr. Trump—were helpful but insufficient.

Some have come to believe that a distrust of institutions is more pervasive than anticipated across many voter groups, and that it leads conservative voters, even those with college degrees and urban addresses, to avoid participating in polls in disproportionate numbers. If so, the problem likely can’t be corrected by adding more members of any one demographic group to a polling sample, they said.

“I readily admit that there were problems this year, but it is too soon to know the extent of the problems, or what caused them,’’ said Courtney Kennedy, who supervises poll methodology for the Pew Research Center. In both 2016 and 2020, she said, ”there was a widespread overstatement of Democratic support,’’ but the causes this time around could be different.

Some pollsters said that state and local polls had been particularly off-course. “District-level polling has rarely led us—or the parties and groups investing in House races—so astray,’’ wrote analyst David Wasserman of the Cook Political Report, whose review of private polling by political groups had led him to predict Democratic gains in the House of between 10 and 15 seats, rather than the GOP gains that now seem likely.

But national polls also faced challenges. The final Journal/NBC News poll, taken in the past week of the campaign, found Mr. Biden leading by 20 points among women and 23 points among seniors—numbers that seemed to require Mr. Trump to put together an improbable coalition of voters to win.

By contrast, a large survey of the 2020 electorate called AP VoteCast found that Mr. Biden won among women by 11 percentage points and that he lost seniors by 3 points, rather than winning the group.

Moreover, the Journal/NBC poll appeared to understate the number of new voters that GOP recruiting efforts were bringing into the electorate, as well as their tilt toward Mr. Trump in some swing states, said Jeff Horwitt, a Democratic pollster who conducts the survey with Republican Bill McInturff.

In other words, Mr. Trump won several key states in the election in part because he changed the electorate, a possibility the pollsters had signaled but which the poll’s results didn’t fully reflect.

“First-time voters are hard to reach for the same reason that you have to devote resources to get them to vote for the first time—they’re harder to engage,’’ said Mr. Horwitt. He added that he wanted to see final, nationwide results to better assess the performance of the poll.

Pollsters were grappling with whether the 2020 results showed new sources of error, or whether problems known from 2016 hadn’t been fixed.

A panel of the polling industry’s professional group, the American Association for Public Opinion Research, said that one source of problems in 2016 was that an unusually large group of voters—some 13% in several swing states—only settled on a candidate in the final days. That meant pollsters missed those voters’ late swing to Mr. Trump.

AP VoteCast, the large survey of the 2020 electorate, found only 5% of voters choosing a candidate in the final days, and they broke more modestly for Mr. Trump.

Another problem the panel identified: Some pollsters of swing states in 2016were using methods that understated the voices of white, working-class voters and likely of rural voters. Some pollsters this year adopted statistical and other measures aimed at correcting the problem, but some didn’t.

Polling continues to face the challenge of declining participation. As fewer people pick up the phone for unknown callers, survey-response rates have fallen from 36% in 1997 to 6% in 2018 at the Pew Research Center, the group says. The rise of cellphones has raised costs, since by law they can’t be called with autodialers, prompting pollsters to hunt for new ways to find respondents.

Mr. Blumenthal and others were skeptical that polling faced a significant problem this year from “shy Trump voters”—the idea that some voters, when interviewed, believe that it is socially unacceptable to say they support Mr. Trump.

The bulk of the problem, he said, is less likely to be that people are misleading pollsters than getting the right mix of people on the phone to begin with. That is likely to lead pollsters to think harder about who isn’t participating in polls, and whether a meaningful number of people across many demographic groups, who in the main are conservative, decline to participate. While many pollsters monitor the share of Republicans and Democrats who participate and decline to participate, “it may be that the Republican voters you interview are on the more moderate side, whereas the ones you didn’t interview might be more staunchly part of the Republican base,’’ said Ms. Kennedy.

It is a hard problem to study, Mr. Blumenthal said. “We don’t interview the people who don’t respond,’’ he said. “We know much less about the voters we cannot reach than the ones we interview.’’

This story has been published from a wire agency feed without modifications to the text