Republicans cleaned up during Tuesday’s election and no one saw it coming. Why?
Since long before 2016, when the utterly unforeseen election of Donald Trump shook news pundits on three continents, polling was starting to go awry.
The equally-surprising passing of Brexit in the U.K. should have tipped U.S. pollsters gearing up for the 2016 presidential election that something was amiss, but it did not.
In the four years after Trump was elected, much was said about Donald Trump. In fact, there isn’t much that hasn’t been said about the former president.
What we never heard, in all the years after Trump was elected, even after Democrats failed to make the kinds of gains pollsters confidently predicted they would in the 2018 mid-term, was how pollsters fixed whatever was broken in 2016.
In 2018, Democrats took what modest gains they could get and didn’t ask too many questions. In 2020, it was the same story; Joe Biden was elected; but he was predicted to carry the race by a much more comfortable margin.
If Democrats recall, most went to bed on election night with the sinking feeling Donald Trump was about to be reelected president.
Down the ballot from Biden in 2020, Democrats lost 27 of 27 toss up races pollsters predicted they would win.
Still no one asked any uncomfortable questions of pollsters.
In all the years since Trump’s upset in 2016, there haven’t been any headlines like, “Pollsters Solve Polling Methodology Crisis!” or “One Polling Company Has Discovered a Way to Get a Random Sample Again,” or, “Polling is about to become a whole lot more reliable, here’s why…”
Now, as Democrats dig out of an emotional week, after an emotional few months, as moderate and progressive Dems battle it out over who is to blame for Virginia and New Jersey, and Republicans level their usual charges of “Democrats in disarray,” nothing is being said about the bad polls.
In all the finger pointing, no one is asking why pollsters keep getting it wrong, and how they plan to make it right.
Pollsters showed New Jersey’s Gov. Phil Murphy with a double-digit lead over his Republican challenger the entire race. Instead the two candidates were neck-and-neck. Pollsters assured Democrats in Virginia that Terry McAuliffe was going to win; McAuliffe led now Governor-Elect Glenn Youngkin in every pre-election poll.
And lost the election.
If the Democratic Party and the Murphy campaign had known how close the race was, if pollsters had been more accurate or more honest about their predictions, would they have campaigned harder?
If Democrats in New Jersey had known Murphy was in serious trouble, not coasting to an easy reelection, would it have increased Democratic turnout?
If Democrats in Virginia had known Glenn Youngkin was actually several points ahead of their Democratic candidate, might the race have played out differently?
It is well past time Democrats, and Republicans, started demanding answers from pollsters who are intentionally or unintentionally misleading the public.
It is time for pollsters to admit the flaws in current polling methods are terminal and tell media companies and political analysts how they plan to fix it.
Not that such an undertaking will be easy. There are several fundamental problems with polling in general.
Polling is Flawed From the Outset
Polling, by its very nature is flawed.
Imagine you are a state fair where 100 people are in attendance. You are taking a poll about something- let’s say, how townspeople feel about adding an extra room to the schoolhouse.
Out of the 100 people, only maybe 20–30 agree to be polled. The rest, the majority, ignore you. For whatever reason, they refuse to be polled. People who refuse to be polled can’t be polled. As a result of this simple truth, polls only ever reflect the opinions of people willing to be polled, which is not a true sample of the population, no matter much pollsters want it to be.
People who enjoy answering in polls, and take the time to do so, tend to have a few traits in common not shared by the populace at large- they tend to be more likely to volunteer, for instance.
People willing to be polled also have a tendency to have strong opinions about the polling subject, one way or the other. This, of course, has a tendency to skew the results, making the polls appear more polarized.
The Cellphone May Have Killed Polling
When we lost the landline, we may have lost a great deal more than a long cord with a tendency to tangle and rotary phones that took ten minutes to dial.
Landline numbers gave pollsters their best method of generating a truly random sample of people. With the death of the landline, and the advent of cellphones, pollsters found other methods.
They have not been as accurate.
Online polls are problematic; not everyone is online. Only about 1 in 5 U.S. adults use Twitter. Email queries are dubious, purchased marketing lists aren’t ideal solutions, either.
Whatever method they use, pollsters must find a way to stop their skewed polls from interfering with the electoral process. This is very important because:
Polls Only Ever Get It Wrong One Way
Polling is skewed. It is has been badly wrong since Brexit and Donald Trump’s surprise victory in 2016. In every election since, pollsters have been way off and their faulty predictions only ever break one way:
In favor of Democrats.
When is the last time you heard of a Republican who was projected to win in the polls going on to lose the race? When is the last time a Democratic candidate had an upset?
It never happens because pollsters are over-sampling Democrats by a wide margin.
Part of this is an unintended consequence of the click-bait era. News networks like CNN, MSNBC, and ABC- and their online equivalents- have seen a sharp decline in viewership over the past few years.
People like to nurse their confirmation biases; when these networks stopped covering Republicans with even a semblance of objectivity, Republicans turned the channel and they haven’t been back since.
Rachel Maddow, Joy Reid, Don Lemon, and other popular “mainstream” news personalities are increasingly broadcasting into an echo chamber of like-minded individuals. When these networks take a poll, chances are good that very few Republicans would even hear about it.
It would be like Tucker Carlson taking a poll of the liberals who watch FOX news. That’s a joke because as everyone knows, liberals don’t watch FOX. As unbelievable as it may be to progressives, who still think they are getting unbiased news from these sources, Republicans feel about CNN, MSNBC, ABC, NBC, etc., the way Democrats feel about FOX.
They don’t believe a word of it.
Republicans also have growing doubts about these bad polls being good faith mistakes by pollsters working out the kinks in new polling methods.
Going back to our survey of 100 townspeople, if a pollster was trying to get a “yes” on adding to the schoolhouse in order to increase support for the project, they could do so in any number of ways.
One way would be to survey a portion of population likely to give a favorable answer, then broadcast that inflated number to bolster popular support for the project.
“See, everyone else is supporting the schoolhouse expansion- 80% of the village according to this poll, why don’t you?”
The “everyone else is doing it so you should, too,” has a certain influence. If the survey showing 80% support was really not a random survey of townspeople, but a survey of 20 locals and 80 schoolteachers from across the country, it isn’t a survey at all; it’s a campaign ad and should be treated as such.
If a demographic is outsize in the sample, pollsters should disclose that fact. Barring that, pollsters must improve their random sampling in order to remain relevant.
Polls have been wildly off in every last major election since 2016. At some point, someone is bound to notice.
(contributing writer, Brooke Bell)