November 3, 2020 is a historic date. Not because of the presidential election, but because it should mark the death of the media’s reliance on public polling.
Regardless of whether Joe Biden or Donald Trump ultimately prevails, the media needs to once again evaluate their behavior leading up to Election Day. For the past several election cycles, the American public has been fed a steady stream of polls, conducted with outdated methodology, to give us a definitive view of who is winning. They were wrong in 2016, they were wrong again in 2018 and they are wrong again in 2020. It has gotten to the point that if a state poll is within the margin of error, it’s more happenstance than any true foresight.
Public polls are like drugs to political reporters and to talking heads, but, most of the time, they do not even understand how to use the data they are shamelessly using to shape their narratives.
While there is a definite science to polling, it is still far more of an art. The outputs are completely dependent upon the inputs, which are a judgement call by the firm conducting the survey. And then, for a basic primer in how it all works, the polling firm has to model the data to determine the percentages of people who will vote. In years past, they could look to prior similar elections to create accurate models. The prevalence of cell phones began to impact accuracy, and then Donald Trump happened.
In 2020, after they had gotten it wrong so many times before, we were still expected to believe that pollsters had an idea of how to model an election in the midst of a global pandemic. Oh, and there was an unprecedented amount of early voting taking place. Did I mention that Donald Trump, the man who defies political norms, was still running?
Yet, the media continued to grasp at every poll issued via press release to spin the flawed numbers. They committed the sin of treating polls of registered voters the same as polls with likely voter models. Worse than that, they put an outsized level of importance on national horse race numbers, diminishing the fact (much to Hillary Clinton’s chagrin) that we elect our presidents via a series of statewide elections.
Individual state elections are even harder to get a sense of the electorate. A week before the election an ABC News/Washington Post poll had Wisconsin going to Biden by 17 points!
There is still value in polling, in that they can tell what messages work to change opinions of segments of the population, or to give a sense of what people are thinking. What they are not good at is predicting outcomes. The fact is, political professionals do not use polling like the media does. And in recent years, even private polling has been a dumpster fire. Sure, we track the horse race as a baseline, but polls are mostly used to test what messages move the electorate.
After 2016, the media did a self-evaluation of why they missed the outcome. They largely decided that they had not paid enough attention to rural America. While that continues to be true, it is also because they all still overly rely on public polls and then talk down to anyone who disputes their efficacy. It’s far safer to be in their echo chambers with people who largely agree that the polls are right.
Public polling should rest in peace, once and for all. However, I suspect some zombie variation will take life in 2022 and the media will again be pitching these relics to the American people.
Jai Chabria is a Managing Director at Mercury, a Global Strategy Group. He is former Senior Advisor to Ohio Governor John Kasich.
