Alan Zendell, September 25, 2023

Voter preference polling is one of the media’s favorite tools for grabbing attention. It plays a vital role in ratings wars and attracting sponsors, but the lingering question remains: are the results meaningful? Public response to polling, especially more than a year out from an election and before a single primary vote has been cast, is mixed. Some people read polls on the edge of their seats while many others ignore them. Sadly, most people don’t understand them well enough to judge.

To people who do understand, particularly in a presidential election, national polls don’t mean much, because in today’s world, a handful of swing states determine the outcome. In 2020, the margins in five swing states (Pennsylvania, Georgia, Wisconsin, Michigan, and Arizona) totaled 280,000 out of 155 million, less than 0.2% of the votes cast. They mean even less the further away the election is.

Some polls, like the just-released ABC-Washington Post survey on the 2024 presidential election, generate considerable controversy. There are many polling organizations, some that purport to be entirely objective, while others have clear biases. Those biases can manifest in the way sampling is conducted and in the way questions are worded. In virtually every poll I’ve participated in, I felt that most of the questions weren’t biased as much as worded in a way that I didn’t have a clear response.

The problem with the Post-ABC poll, which was conducted from September 15-20, was that its results were very different from polls conducted by other groups. The Washington Post acknowledged that simply by describing it as an outlier – a correct conclusion, but one that caused me to ignore it completely. Even if it had agreed with other, similar polls, I’d have had trouble taking it seriously. Why?

Consider how polls are conducted today. According to The Post, it used “a random national sample of 1,006 U.S. adults, with 75 percent reached on cellphones and 25 percent on landlines,” and the results have a margin of error of at least 3.5%, 4% among registered voters, and it’s much larger than that when results for smaller subgroups are presented. Four percent doesn’t sound like a lot, but that number is only valid if all the assumptions made in conducting the poll were correct.

The most important assumption is that the sampling universe accurately reflects the opinions of likely voters. One problem is that 11% of the people sampled for the ABC-Post poll were not even registered to vote. But the more serious problem is whether even the 89% who were registered actually represented the nation at large.

The Post reported that about 750 people were contacted on cellphones and roughly 250 on landline phones. Thus, the entire poll was conducted by interviewing people who answered their phones and were willing to speak to pollsters. And for people who work for a living, that means they could only be contacted outside their work hours. Does that sound like anyone you know is represented by the sample of people questioned? I and virtually everyone I know ignore phone calls from numbers I don’t recognize or those identified as likely spam or marketing calls. So, who is actually being counted in the results?

It sounds to me like the only people who respond are those who answer their phones every time it rings. Maybe they’re lonely or bored or have nothing better to do, given that more than nine times out of ten, the caller is either a robot or someone trying to sell you something you neither need nor want. But that group certainly doesn’t think the way I or most of the people I know do. Another factor is that even among those people who answer a pollster’s call, most are too busy or disinterested to take the time. My guess is that people who are willing to spend fifteen minutes talking to pollsters have intense feelings, usually including anger bordering on rage about one of the candidates or a hot-button policy issue.

I cannot take such polling seriously. I have no confidence that either the sampling universe used represents actual voters who enter a polling place or that the questions they posed really touch on what most voters think is important. As RCA Chairman David Sarnoff, in the movie Twenty One, says to the federal investigator who proved that the quiz show of the same name was rigged – No one ever said it was honest. It’s just entertainment.

I recommend that we view national polling the same way. State by state polls, for either primary or general elections, especially close to voting days are likely to be a lot more reliable, but in the end, it really doesn’t matter what anyone else thinks. Your vote belongs to you.