Fool me once, shame on you. Fool me twice, shame on me.
And there’s plenty of shame to go around in this post-election postmortem — especially for those who thought the polls predicted either a red mirage or a blue wave of victory.
Easy to get fooled when several polls said Texas was a dead heat, and even a possible flip to blue for former Vice President Joe Biden. Others claimed he’d possibly win traditionally red Iowa, and many cited stats showing President Donald Trump was significantly behind in Florida. All of those were wrong, and some — like the Florida surveys — were way off. On election night, Trump had an early wide lead in his adopted home state.
I’ve always been suspicious of political polls. That distrust was underscored after the 2016 presidential race, when the vote essentially turned predictions upside down. Now the pollsters will tell you that the final tally from four years ago did, in fact, prove them right; Secretary Hillary Clinton won the popular vote, as they said, topping Trump by 3 million ballots.
But I would argue if you are like the average news consumer, you heard the constant polling drumbeat of her potential for victory as confirmation of victory.
To be fair, there was frequent commentary about Clinton’s need to win the crucial Electoral College map. But expert pollsters told a twin tale, urging caution while at the same time pointing to past scenarios that seemed to indicate a win — information that the ubiquitous pundits amplified ad nauseum.
I believe in data, but it’s clear that the science of gathering an accurate sense of Americans’ voting patterns is outdated, or at the very least critically flawed. Maybe the sample sizes are too small or too city-centered, or perhaps it’s still hard to surface voters through cell phone outreach. Bottom line: the survey sampling is not broad or deep enough.
In 2016, I bought the idea of large, hidden groups of so-called “shy” Trump voters who didn’t want to publicly state their support for the brash candidate. But in 2020, there was no shortage of eager and enthusiastic supporters of the president who were loud and proud about voting for him again. Why weren’t more of their huge ranks reflected in the surveys? I’m convinced it all comes down to human behavior, which psychologists and behaviorists will tell you is often unpredictable. There’s simply no amount of number crunching, algorithm juggling, and the best technology money can buy that can produce absolute results.
Like the human behavior I questioned David Plouffe about during the 2016 campaign. Plouffe was campaign manager for Sen. Barack Obama’s successful 2008 campaign, and later a senior advisor to the president. He was an executive at Uber when I interviewed him. It was August, and then candidate Donald Trump was barnstorming the country as the Republican nominee. Enthusiastic voters were driving hours to his rallies, standing outside when the inside venues filled up. The polls all said Secretary Hillary Clinton had a strong lead.
'But what about those rally attendees?' I asked him. 'Couldn’t they indicate a huge voter turnout for Trump?'
'Nope,' he said. 'They don’t vote.'
I pushed back. 'Who drives hours to attend a candidate’s rally, if they don’t intend to vote?'
He explained it was more complicated than that, and I’d see.
Well, I did see. And the results then and now prove what the veteran politicos always say: The only poll that counts is the one on Election Day.