Alan Zendell, June 15, 2024

I’ve talked about political polls before, but until Americans understand the way we’re being flummoxed by the media, I’ll continue to talk about them. With four-and-a-half months before the November election, the danger posed by inaccurate polls is extreme, because this generation of Americans is shockingly gullible. If the media consistently tell them candidate A is ahead, that tends to become a self-fulfilling prophesy over time.

In order for polls to be accurate predictors, the people polled must be representative of the voting population. The purpose of polling is to get a sense of how the larger population might vote based on interviewing a small sample of prospective voters. I won’t go into the details of sampling and error rates – all you need to know for this conversation is that statistical analysis works when it’s done right. You can trust that a properly designed sample of a thousand interviewees can tell us how millions of voters are likely to cast their ballots with a likely error of about three or four percent – the key being “properly designed.” The problem with the way predictive polls are being conducted today is that the people polled cannot possibly be a representative sample of the electorate.

To provide context, I’ll discuss polls that can usually be relied on to yield meaningful results. People who analyze how voting trends change over time rely heavily on exit polls from previous elections. An exit poll is exactly what it sounds like. Interviewers intercept people exiting their voting locations and ask who they voted for and why. I tend to trust these polls far more than predictive polls because the only obvious flaw in the sampling is that everyone they approach may not want to be interviewed.

Most pollsters are not politically motivated, though the people who hire them and broadcast their results usually are. Moreover, all professional pollsters understand how to create statistically valid sampling universes. It’s a lot easier than brain surgery and almost as easy as rocket science. But knowing how is not the same as doing it. Exit polls work because people are more likely to accurately tell you who they voted for than who they plan to vote for, and there’s no reason to think that people who are willing to talk to exit pollsters vote any differently on the average than people who refuse to.

When your favorite news network tells you that more people under 45 voted for Trump than Biden in 2020 while the reverse was true for people older than 45, you can believe that with a reasonable degree of certainty. But what happens if the pollsters try to create a sampling universe to ask a similarly diverse group of people who they plan to vote for next November? I assert that it can’t be done using today’s polling methods and technology, and every network that broadcasts poll results knows that. But polls attract viewers, be they knowledgeable or naïve, and sponsors who pay the bills like to attract viewers. It’s very much in their self-interest to treat elections like athletic competitions whether the results are meaningful or not.

In an exit poll, the sample universe is well-defined – it’s all the people streaming out of their voting booths. All the interviewer has to do is find people willing to talk to them in as unbiased a manner as possible. But creating a sampling universe for predictive polling is very different, and in today’s world, virtually impossible. Today’s pollsters rely almost entirely on cell phone interviews and email questionnaires. Ask yourself if you or anyone you know is likely to participate in them. To do so, they’d have to be willing to answer calls from unknown callers or respond to emails from unknown sources who are most likely trolling for contributions.

Maybe you have a 95-year-old grandparent who’s bored and lonely enough to talk to anyone who reaches out to them, but who else do you know who’s likely to? As I’ve asked in previous articles, if pollsters aren’t talking to the vast majority of us who screen our calls and emails, who, exactly, are they polling? Whoever they are, they can’t possibly be representative of the voting population.

The real villains here are the giant media companies. Have you ever heard a polling “expert” cast doubt on the numbers they tout? They don’t care if the polls are accurate as long as people tune in to see them, and that’s incredibly dangerous for a generation of Americans that has largely forgotten how to think for themselves. Between inaccurate polls, internet bots, and virtually unregulated platforms like Facebook and X (Twitter) most Americans have no idea whether what they see and hear is truth or fiction. This year, polls that continually predict Trump leading, despite every logical argument that says he shouldn’t be, are a major part of the problem. But it’s a problem each of us can solve by simply ignoring them and deciding for ourselves.