The American people may be politically divided, and the emerging 2020 election results certainly reflect that reality. But I suspect we all could agree on one thing: pollsters who were trying to take the temperature of the American voter during this election cycle didn’t exactly cover themselves with glory. To the contrary, the polls were remarkably, and dramatically, inaccurate predictors of the actual results.
This year, virtually every pollster produced results that diverged from reality by percentages that far exceeded the “margin of error.” Florida, which was one of the most highly sampled states in the country, provides a good example of the phenomenon. The final pre-election polls, by respected pollsters, predicted that Joe Biden would win the presidential election by between four and six points, and the final results had President Trump winning by about three and a half points. That means the polls that were announced with great fanfare and breathlessly discussed by talking heads on news shows were off by between eight and ten percentage points. And Florida is not alone. Across the country, in the presidential, Senate, and House races, the vast majority of the polls were simply wrong.
Why were the 2020 polls so wrong — even after pollsters vowed that they had learned their lessons and tweaked their procedures in the wake of 2016, when polls also were demonstrably inaccurate? People have come up with a lot of theories. Maybe the pollsters aren’t very adept at predicting who is actually going to cast their ballot and are sampling the wrong populations. Maybe the polling questions reflect intrinsic bias. Maybe there are “shy” voters out there who don’t want to admit who they really support. And maybe people don’t like having their days interrupted by intrusive pollsters, and are increasingly likely to lie about their true intentions and feelings as part of an effort to consciously mislead the pollsters. Or maybe, just maybe, the notion that polling can be viewed as reasonably “scientific “ is a charade, and we should just accept that trying to detect and predict political currents in a country as broad and diverse as the United States is a fool’s errand.
I personally think we’d all be better off if there were no publicly announced polling. Polls don’t advance the national discourse, and they have made journalists into lazy, incurious creatures who don’t venture outside to actually talk to real people or cover real issues. If reporters spent less time trying to analyze inaccurate polls, they’d have more time to actually do their jobs. And if there is any chance that poll results cause certain voters not to exercise their franchise — either because the polls show their candidates to be far ahead or far behind — eliminating polls would eliminate that vote-suppression factor. We can also, I think, agree on the proposition that anything that suppresses voting is not to be encouraged.
I hope people remember the inaccuracy of polls when the next election rolls around, but I also hope the news media does some soul-searching about how it covers poll results. This election cycle demonstrates that poll results aren’t really news in any meaningful sense, and shouldn’t be reported as such. When polls are off by double or triple the claimed margin of error, they are little more than speculation, and not much more credible or informative than reporting on the armchair predictions of your relatives and friends. A case can now be made that, if the news media really wants to stick to reporting news, it won’t report on poll results at all.