The future of polling may be on the ballot this year.
POLITICO might have said it most eloquently, though everyone else in politics is saying it, too: “Pollsters sweat another Election Day reckoning.”
Have pollsters fixed the problems with inaccurate polling?
And if they haven’t, have pollsters polled their last?
Polling, which failed miserably to predict outcomes in 2016 and again in 2020 — and since during subsequent governor’s races in Virginia and New Jersey — is getting another shot this year.
“Final Senate Forecast Map Predicts Who Will Control Congress After Midterms,” reported Darragh Roche for Newsweek on Election Day, 2022.
While POLITICO maintains that “Independent polls are painting a very different picture of the midterm election than the Republican-aligned firms pointing to a GOP victory,” most pollsters, including 538’s Nate Silver, believe polls are still probably under-sampling Republican voters this Election Day.
Most of the final polls, betting markets, and other prognostications favored Republicans in the election's closing days.
In the House of Representatives, Democratic Party hopes have gradually been fading over the past months. The Senate is another story. Until the final closing days of this election cycle, many Democrats in leadership were still bullish about keeping the Senate majority, if only barely.
As Election Day approached, the bottom seemed to fall out of that hope. As voters headed to the ballot box Tuesday morning, Republicans were suddenly favored to pick up the Senate as well as the House. Races that were predicted to be toss-ups, including the Governor’s race in Kansas, were suddenly moved to “leans Republican.”
The Georgia Senate race, which Democrats thought they had in the bag three months ago, is instead going to be a down-to-the-wire nail-biter. Republican challenger Herschel Walker, who the DNCC viewed as no threat whatsoever to Democratic Party incumbent Sen. Raphael Warnock until this month, has now caught up in the polls.
Some polls show Walker ahead. If Georgia’s vote is closely split between the two candidates — as is likely — the race will become a runoff, in which case it may be another month before the American people know who prevailed or which party will control the Senate in 2023.
Across the nation in deep blue areas, Democrats have been flagging badly in the polls. Oregon, New York, Rhode Island; Republicans are performing better in the polls than they have in decades. Some are within the margins of error for beating their Democratic opponents; some are even closer.
In Pennsylvania, Republican Dr. Mehmet Oz has been steadily gaining on Democrat John Fetterman since the latter’s disastrous debate performance in October and, as of the early hours of November 8, is now favored by 538 to win.
The best, most experienced pollsters still have major questions, even as the voters cast their ballots. Many will openly admit they have no idea what will happen once the polls close and the votes are counted.
How valuable polling will be going forward may depend on whether or not political parties and media companies need a weatherman who can predict the weather, but only once it’s happened.
“The 3 Big Questions,” Nate Silver of 538 still had about Election Day on November 7, weren’t piddling matters. “Will the polls be systematically wrong?” was Silver’s, “Question 1”.
He didn’t add the qualifier “again”, but it is implied. Pollsters, intentionally or accidentally, have been missing the mark in the last few election cycles. Silver’s other big questions, “How big will the turnout gap be?” and, “How much does candidate quality matter?” weren’t exactly non-issues either.
Some of the reasons for the failure of polling post-2016 are endemic to polling itself and intractable.
Political polls aren’t a sample of likely voters. Polls can only sample people who are willing to be polled. Not everyone is willing to be polled. Many people are adverse. One of the reasons pollsters keep giving for their failures is that “response rates suck.”
People don’t answer unknown numbers anymore. Getting people to respond to a long survey online is much easier but not everyone is active online. Many of those who spend a great deal of time online tend to fall within certain generalized demographic qualifiers; younger voters are more likely to be spending more time online than their octogenarian counterparts. White-collar workers tend to spend more time online than working-class people with blue collars, etc.
Polls also oversample people with strong opinions on the subject being polled. Strong opinions, for or against, tend to make people more willing to sound off about a particular topic, policy, or politician.
This phenomenon ignores wide swaths of likely voters who may not have very strong opinions on things like abortion and Ukraine, to say nothing of the “death of democracy.”
Not everyone pays attention to politics at all. Some voters might not have heard democracy was ill.
Even pollsters of Silver’s quality have been reduced to wondering things like, “if Republicans beat their polling averages by 3,”, “if Democrats beat their polls by 3 points across the board”, and ultimately, “Which type of polling error is more likely?”
What is to become of polling?
That pollsters, not to mention voters, may not even have all the answers on election night — and why that particular phenomenon, which seemed to be a feature of COVID-era voting, seems to be sticking around — are other factors that hallmark the changing the face of politics in America.
Slow-rolling the results of the election undermines faith in the election process. If anything, counting votes should be easier and faster than ever before given various improvements in technology.
Polling that is continuously wrong also undermines faith in the election process. Increasingly, polling is being seen as a tool for influencing public opinion, rather than a tool for reflecting it.
“So what will it mean if Tuesday is a GOP rout that nets the party dozens of House seats and easily delivers the Senate majority?” POLITICO asked one of their favored pollsters on the eve of Election Day.
“At the simplest level, it means nothing we’ve done since 2016 has fixed the problem,” answered Marquette’s Charles Franklin.
(contributing writer, Brooke Bell)