Polls have been consistently wrong in one direction during the last few election cycles. The problems have NOT been fixed.
Something is rotten in the world of polling and it isn’t clear to industry experts how to fix it. Some doubt whether or not the problems inherent to modern polling methods are fixable.
Once the landline was replaced by the internet, it became much more difficult for pollsters, politicians and media companies to reach a truly random sampling of people.
Online polling hasn’t exactly proven a time-tested method. Not everyone spends the same amount of time on the internet and in online forums.
While 100% of journalists at the New York Times have active Twitter accounts, only about 22% of average Americans can say the same. Even the editor of the NYT is starting to warn employees about the dangers of over-reliance on their Twitter “echo chambers”.
Online polls in general tend to favor the college-educated, younger, white-collar, upper middle-class demographic. This demographic often spends a great deal of time online unlike some of their working-class counterparts. This demographic also largely skews heavily towards the Democratic Party.
On the other hand, there are some people who barely spend any time online at all, having little to do with social media, online news, email, or any other digital avenue by which pollsters could reach them. Many members of the working class falls into the latter group. Older people often spend less time on the internet and on social media than their younger counterparts. Disadvantaged and impoverished people also tend to spend less time on Twitter than their wealthier counterparts.
There are other reasons online polling isn’t the answer: The Boaty McBoatface Principle.
Online polling is little more than an online contest, like the naming contests that have frequently gone amiss when some online prankster hijacks the contest to intentionally skew the results. Like voting to name a high-tech scientific maritime vessel “Boaty McBoatface” or a popular whale at a local aquarium “Mr. Splashy Pants.”
Online polling contests gone awry are responsible for banishing rapper Pit Bull to perform in Alaska and favoring a school for the deaf in a contest for a Taylor Swift concert.
Online polls are untrustworthy.
Reaching people in the real world is well-nigh impossible.
A random sampling from the phone book was once enough to give pollsters a good bead on things. Now, pollsters can reach all the online respondents they could possibly want- cheap.
Reaching those working-class voters who eschew the internet to get their opinion is expensive and extremely difficult. To reach these under-sampled demographics, pollsters have to call, and call, and call, and call.
Over a dozen calls can occasionally produce a response but it is easy to see how pollsters and media companies would take the simpler way out: Using online polling for the majority of their responses and heavily weighting the scant handful of under-sampled opinions to compensate for the difference.
One problem with that: It isn’t working.
Worse, for the Democratic Party heading into a difficult mid-term anyway, is that it is working against the Democrats.
Republican candidates who aren’t projected to win, do sometimes still win. When Republicans don’t win, they don’t lose by as much as pollsters predicted- not anymore. When Republicans are projected to win, they win.
Democratic Party candidates projected to win sometimes do win; but sometimes they lose. Democratic Party candidates projected to lose, often lose by much more than was projected. Democratic candidates predicted to lose always lose.
When is the last time Democrats pulled an upset at the ballot box?
For Republicans, it was 2021.
Those wonderful polls predicting Hillary Clinton’s runaway victory in 2016, the previous upset vote by Britons for Brexit; polls predicting Trump’s sound defeat in 2020, rather than the hair’s breadth victory President Biden and the Democrats managed to eke out in the midst of a pandemic. Pollsters also missed the shellacking down-ballot Democrats were dealt in 2020.
In the glow of ousting Trump, and in cinching those two Georgia Senate seats, plenty of Democrats missed the fine print where pollsters admitted they called 27 out of 27 toss-up districts for Democrats that went to Republicans instead. That isn’t exactly batting a thousand.
All these missed predictions are like candy: Delicious for Democratic voters in the moment it takes to read about a poll favoring Democrats to win, nutritionally empty promises revealed in painfully unexpected defeats on election day and uncomfortably close, costly races where there should be smooth sailing.
But what on earth is suddenly happening? Suddenly, Republican candidates are leading in the polls- even in places they have trailed election cycle after election cycle.
Is this a sign that polling problems have now been fixed and media companies are now reaching an equal number of Republicans and Democrats in a representative sample of likely mid-term voters?
There have been no major breakthroughs in polling. No glowing bio pieces about the pollster or polling company who finally cracked the code have appeared. There haven’t been any cerebral think pieces about how cleverly the puzzle was solved. In fact, during the last major election, pollsters struck out yet again.
Republican Glenn Youngkin, against all polls and political oddsmakers, was elected Governor of Virginia not long ago. In nearby New Jersey, a Democratic Party stronghold, incumbent governor Phil Murphy, in spite of polls showing him with a very comfortable lead the entire race, came this close to being unseated by a Republican challenger.
In the months since those elections, there have still been no stories about polling being fixed, written by media companies, pollsters, polling companies or anyone else. No one has stepped forward to say: “Sorry we’ve been getting it wrong, we’ve fixed it and November 2022 will be different.”
At the heart of these polling reliability issues is the inability of polling companies and mainstream media outlets to reach conservative and even moderate-leaning voters, including wide swaths of the working class.
With trust in media companies depressingly low, the problem has only grown worse. A not insignificant percentage of Americans are openly antagonistic to corporate media outlets and wouldn’t participate in a poll if you paid them.
If anything, some disgruntled voters are more likely than ever to intentionally mislead pollsters.
Polling is looking more unreliable than ever heading into November. As polls are still heavily favoring Democrats, it is worth considering that the news might actually be worse for the party that triumphed in 2020.
If pollsters strike out again in 2022, perhaps America could ask them a question: How many elections do pollsters fail to predict before polling is declared defunct?
At some point, it begins to look like a vanity project: All shadow during campaign season, no substance on election day.
(contributing writer, Brooke Bell)