Political polling has existed in presidential elections for over a century and the industry has never been more dominant, with major poll operators serving party and candidate clients, as well as broadcast and cable networks. Polling gets things right much of the time — but when they are wrong it gets attention. This year, the damage is still being assessed — but much like the election in 2016, it appears the issue with polling is a combination of perception and misinterpretation.
The three broadcast networks and three cable news networks each have their own polling partners, who are often leading newspapers and magazines. The biggest problem, as a result, is how central the polls have become to most political news coverage as the election season unfolds.
In 2016, the national popular vote polls actually ended up fine. The consensus lead on election day for Hillary Clinton was about 3 percent, and she ended up leading by 2 percent. But most key state polls were off when compared to who won. And the earlier perception bestowed by the media of a greater Clinton lead — as well as an inherent sense that Donald Trump couldn’t win — led to a number of incorrect assumptions in coverage.
This year, it is way too early to gauge the popular vote errors. The end of race average had Joe Biden up eight points; some analysts suggested his margin of error could be in the five to six point range. In the end, if the results hold as projected, his results would be close to the margin of error.
State polls — again, with the caveat that final results are not yet known — actually averaged results that in many cases of swing states correctly indicated the winner. Often — like in Florida, Georgia, and North Carolina — they were realistic in their “go either way” prognoses. But anyone following closely by Tuesday had reason based on polls and the resulting news coverage to believe that Michigan and Wisconsin were safe Biden states, for example, not the nailbiters nearly as close as 2016 that they turned out to be.
Worse were the polls of most Senate races. What were thought to be toss-ups or Democratic candidate leads in states like South Carolina, Iowa, and Montana turned out to be fairly easy Republican wins, and undercut the long media narrative that a blue wave would change control of the Senate.
The overstating of Democratic chances at both levels now is a burden for Biden and his supporters. Even if he is sworn in on January 20, the sense is of a party that lost, rather than a candidate who defeated an incumbent president for the first time in seven elections.
Since what they sell is accuracy, big polling errors get major attention. In 1936, the Literary Digest, which had run unscientific polls for several elections, predicted Republican Alf Landon would defeat Franklin Roosevelt’s try for a second term in a landslide. Landon ended up winning two states, and the magazine went out of business within two years.
In 1948, all leading pollsters made a huge mistake. Thomas Dewey had such a big lead weeks out they stopped checking, but were considered the key reason that Harry Truman’s surprise win was such a shock. How they survived can serve as a lesson for now. The 1948 debacle resulted in a nascent cottage industry — often more involved with radio ratings and business clients — into improving their methodology and salvaging their credibility for political polling.
It is obvious the industry will need to do so some improvements again, and figure out how they again missed a clearly large minority of American voters who both stand in lockstep with Trump but also seem to be less visible when polled.
And the soul-searching should be extended to those who utilize these polls. Anyone who watches presidential races — which never end, as the 2024 one starts today, at least on the Republican side — on TV knows that their coverage centers on the horse race. That means polling — often touting their own, even if it could be an outlier — is what is emphasized. Issues and platforms are of less importance than who is up or down.
This year, ironically, despite polling consistency and a sense of a clear winner ahead, the idea that the election was in doubt actually seemed to be overstated. It was a complicated year with a big jump in mail-in voting. The coverage last night found all networks needing to explain inconsistent and often misleading results. As a long time enthusiastic observer, I felt overwhelmed in the early hours with the lack of clarity and certainty. (That actually turned out to be a not bad reading of the events.)
Cable news in particular lives for elections, particularly presidential ones. CNN, Fox News, and MSNBC for the whole year have covered mainly two stories — the election and COVID-19, with the last more downplayed on Fox. (And there is a possibility that the Trump surge can be attributed to endorsing his anti-mask sentiment.) It is part of a bigger problem. We are far removed from the days when Walter Cronkite and Huntley-Brinkley had an evening half hour slot in which a wide range of stories, domestic and foreign, were covered. The competition for viewers means selection mainly of big stories with storylines to hook them.
And as big as any hook for viewer has been polling acumen. It won’t go away. But at a minimum some reorganizing of emphasis and lessening the hype over them is due.