Can we trust the polls?

In a gesture that is part wishful thinking and part reading the polls, CNN has just declared Joe Biden the winner of the election. We can all go home now:

Biden crosses 270 threshold in CNN’s Electoral College outlook for first time

CNN, the de facto media wing of the Democratic Party, is indeed engaging in some hopeful projection here. Every CNN correspondent hates Donald Trump. The only real question here is: What will Jim Acosta talk about if Trump does, in fact, lose the election? What will inspire his trademark smirks and scowls, if his boogyman leaves the White House?

The polls, as of today, are giving Joe Biden a 16-point lead. As we know from previous elections, the polls are sometimes wrong. And I’m not only referring to 2016.

I recall how, in the summer of 1988, the polls projected a blowout victory for Democratic candidate Michael Dukakis. You’ve never heard of the Michael Dukakis administration because there was none. George H.W. Bush skunked him on Election Day 1988.

Even in 1988, the polls were unreliable. Polling, moreover, was much easier then. In 1988, every American still had a landline, and most Americans still answered their phones. There was no reason not to. Caller ID technologies weren’t common on home phones. Random telemarketing calls at dinnertime, while not unknown in the 1980s, did not become a persistent problem until the 1990s. Most phone calls you received were legitimate, and you didn’t know who the person on the other end of the line might be, or what they might be calling about. Maybe something important. So you answered your phone.

Nowadays, of course, most of us rely on cell phones more than land lines—even 50-something dinosaurs like me. For that matter, my 70-something dad is now more reliant on his iPhone than his landline.

We also now assume that most phone calls aren’t important or legitimate, because well, most of them aren’t. I have received as many as six telemarketing calls in a single day, peddling everything from auto insurance to pain management solutions. Unless I recognize the number calling me (and in 2020, every telephone has a caller ID function), I don’t answer. I know I’m not the only one who lets most calls of unknown origin go to voicemail.

And yet, “the polls” still rely heavily on telephone surveys. Others rely on various online polling devices, which are all opt-in.

In other words, “the polls” are going to disproportionately measure the sentiments of those who a.) feel very passionate about the election, and want to talk about it, and/or b.) those with a lot of time on their hands.

I am quite passionate about the outcome of the upcoming election. Nevertheless, my opinion is unlikely to show up in the results of any poll, because I’m unlikely to give any pollster five minutes of my time. Once again, I doubt I’m the only one. The only poll I plan to participate in is the one at my local board of elections.

Some suspicious minds have suggested that the polls have a deliberate liberal bias. Perhaps. But I doubt that liberal media sources would want to give Biden voters unwarranted confidence in the election’s outcome. Even Jim Acosta is sharper than that.

The problem, rather, is the inherent difficulty of accurately measuring the opinions of 331 million Americans, while relying on any form of opt-in method.


None of this means that Biden won’t beat Trump in a 16-point landslide, or that Trump will get the 16-point victory instead. It would be a bad idea to automatically believe the polls; it would also be a bad idea to automatically assume that the polls are wrong in the opposite direction, to an equal or greater magnitude.

The point is: we simply don’t know. This means that it is still very much worth your time to vote on or before Election Day. That’s what I plan to do, the polls be damned.