Polls Showing Biden Trailing Trump Don’t Mean Much Yet. By Jonathan Bernstein, Columnist

Simple approval ratings tell us what we need to know.

November 21, 2023 at 12:00 PM UTC

Jonathan Bernstein is a Bloomberg Opinion columnist covering politics. A former professor of political science at the University of Texas at San Antonio, he wrote A Plain Blog About Politics.

With many Democrats freaking out about polling that shows President Joe Biden trailing or at best even with former President Donald Trump, it’s worth taking a step back and considering what’s meaningful about polls nearly a year before Election Day — and why they have so little predictive value.

Such a step back is especially useful in considering the bizarre findings of a New York Times/Siena survey which, among other things, found a group of voters who prefer Trump to Biden — but prefer Vice President Kamala Harris to both of them. How does that make sense, especially when most polls show she’s slightly less popular than Biden? 1

The main thing to take away from all these polls is simple: Biden is an unpopular president and we don’t need head-to-head election polls to see that. Simple approval ratings, which show Biden hovering at around 40%, tell us what we need to know. But the early election polling certainly confirms Biden’s current trouble.

Political scientists have long shown statistically and veteran reporters know that early polling doesn’t predict eventual results. In fact, until 300 days before the election, polls have no predictive value at all. As of Tuesday, there are still 350 days until Election Day Plenty of examples back up that result, but perhaps the most dramatic is that in February 1984 — a lot closer to the election than we are now — a Gallup poll showed Walter Mondale with a solid 6-percentage-point lead over Ronald Reagan, who went on to beat him by a whopping 18 percentage points. There are examples in the other direction as well, including a 19% lead for George H.W. Bush over Bill Clinton in March 1992.

The explanation isn’t that people change their minds. What’s actually happening is that a lot of voters simply haven’t given much thought to the far-off decision they’ll have to make in November 2024, but some of them will nevertheless answer a pollster’s question about it. We should read these less as tentative early decisions and more as attitudes — usually about the president — based on the latest news events. As those events fade and are replaced by new ones, those attitudes may change. Then, when the campaign starts in earnest, after the national conventions next summer, voters will start to compare the choices in front of them and we can start reading their responses to pollsters as still-tentative, but somewhat more real.

Why and how that happens can be seen in the extended interviews the Times did with those supposed Harris→Trump→Biden voters. Social media was buzzing over the interview with an abortion-rights supporter who opposed Biden because she assumed he was to blame for the state abortion bans that had gone into effect during his presidency. It’s easy to berate such voters, but that instinct is wrong. People lead busy lives and there’s no reason that keeping themselves informed about current events and politics in the space between elections should be a priority.2 The odds are high that anyone who wants to vote on the basis of abortion will make the appropriate connection based on their beliefs by the time Election Day comes around.

Indeed, while campaigns traditionally are unable to do much in the way of persuading voters, what electioneering is extremely effective at is pushing voters to where they “should” be given their underlying partisanship, policy preferences, sense of how things are going and then supplying them with reasons for their decision. That’s why the interviews with these Harris voters were so fascinating. As Times reporters Claire Cain Miller and Nate Cohn say, talking about the election rather than just answering pollster questions comes at least a little bit closer to simulating the effects of a campaign.

And so it’s telling that in those interviews, about half of this handful of voters wind up flipping and saying they would probably, if reluctantly, support Biden over Trump, in large part because thinking it over removes the question from the realm of attitudes toward Biden and toward the question of comparing Biden and Trump. As pollster Natalie Jackson reminds us, “not everyone spends their time obsessing over Trump & Biden.” But that doesn’t mean they have no underlying views about politicians and the parties that can be activated when they focus on them.

Remember, too, that we’re talking here about the subset of voters who pay the least attention to politics. The overwhelming majority of those who are engaged all of the time are solid partisans who are sure to turn out to vote and will vote for their party’s candidate. What moves the polls for the most part are those less engaged, who can be swayed by current circumstances and who may not vote at all.3

None of this means that Biden is in good shape. If the main thing people are thinking about as the election approaches next summer and fall is that times are bad, there’s a good chance they’ll hold it against him, even if they are wary of Trump. It just means that we should take all of the polling now as a response to what’s happening now and how people feel about it — not as a prediction of next November’s outcome.