close
close

Four things to remember about post-debate polls

Four things to remember about post-debate polls

For people closely following the 2024 presidential election, it was inevitable that the conclusion of Thursday night’s debate would spark curiosity about the possible implications for the race.

Whatever happened in the debate would have been the case; debates are one of the few scheduled events that can change the course of the competition. But then, of course, President Biden gave a shaky performance that amplified questions about his age, the central weakness of his candidacy. The question of impact required an immediate answer given calls for Biden to step aside.

There were polls that immediately confirmed the obvious: that Donald Trump had done better over the course of the meeting. But Biden supporters interested in that can find solace elsewhere, in a televised focus group showing gains for the incumbent president, or in opinion polls showing Biden doing as well as other well-known Democrats.

But these kinds of temperature readings are imprecise, because they either don’t measure the actual presidential race or they don’t measure the race after the debate. Given the interest in understanding what the effects of the debate might be, it’s worth articulating what observers might expect.

1. It would take a while before any movement was seen.

There is a strange inconsistency in the way many Americans view the polls. They expect—thanks in part to often unscrupulous pollsters—to get instantaneous assessments of major events, while they mock the polls they dislike by noting how unlikely it is that anyone would take polls.

There is some truth in that last point. Modern polls do not rely on people picking up landlines, a criticism that survives despite having been addressed years ago. Modern polls use a variety of techniques to reach respondents, but they are slow and time-consuming. In other words, you should not expect a series of quality assessments of an event immediately after it has occurred, simply because it takes time to make those assessments.

That sits alongside the other obvious point: that the effects of an event are not all immediate. It takes time for something like a poor debate performance (or the perception thereof) to trickle out, even in the age of the Internet. The perceptions of the debate discussed Thursday night are unlikely to reflect the perceptions of a few days later.

Moreover, using individual polls as assessments of a major event means you’re relying on a single pollster’s judgment about who’s likely to vote or how to interpret the results. Poll averages often provide a more accurate assessment of the state of an election — but that accuracy necessarily requires more polls. And that means more time.

2. Chances are there won’t be much movement.

The Washington Post has its own polling average. Since the beginning of the year, Trump has shown a national lead of about a point… consistently. Our methodology differs from others that have shown more volatility, such as 538’s. But even 538’s has been quite stable, with the average ranging from a Trump lead of around two points to a Biden lead of less than one point .

Why such stability? Partly because the 2024 race is a rematch of 2020. In a normal presidential year, voters learn something about the candidates, whether it’s how Joe Biden differed from his boss when he was vice president (2020) or Donald Trump’s rise to the political appeared on the scene (2016) or people hearing the names Mitt Romney (2012) or Barack Obama (2008) for the first time.

As for the specific question of Biden’s fitness to run, the central question raised by his debate performance, it’s also true that this is already baked into his support to some extent. In a New York Times-Siena College poll released shortly before the debate, nearly half of Democrats said they thought Biden was too old to serve effectively as president. CBS News-YouGov polling released this weekend — and conducted after the debate — found that 46 percent of Democrats say he should not run. Back in February, 36 percent said that. Democrats overwhelmingly supported him for president anyway.

3. Biden’s position is unusually well protected.

That’s mainly because he’s running against Trump.

In a story by The Post’s Michael Scherer on Monday, an anonymous pollster for Biden’s campaign brushed off questions about Biden’s fitness as possibly not being a “measure that voters will use to determine their choice.” It seems like a cynical and selfish observation, but that doesn’t make it wrong.

The 2016 elections, which still seem to be the best benchmark for 2024, show why this is so.

That race offered the country something relatively unusual: two major-party candidates viewed negatively by most Americans. Republicans hated Hillary Clinton; Democrats were shocked by Donald Trump’s candidacy.

Clinton seemed likely to win. After all, exit polls showed that 61 percent of voters saw Trump as unfit for the presidency. A majority of voters saw Clinton as qualified.

Only a fifth of those who considered Trump unfit I voted for him anyway. Why? Because he wasn’t Clinton. One in ten voters who voted in 2016 thought Trump was unfit to be president and voted for Trump to be president. One in eight voted centrally for Trump because they did not like Clinton. About a fifth of all voters viewed both Trump and Clinton unfavorably; Trump won those voters by 17 points, according to exit polls.

The parallel with 2024 should be obvious: Maybe Democrats (and Americans in general) think Biden shouldn’t run, and believe he can’t serve effectively. But he’s not Trump, just as Trump wasn’t Clinton.

It also works the other way around. That same Times-Siena poll found that 1 in 6 Trump supporters — likely voters who say they plan to vote for him in November — think he has committed a serious federal crime. But hey! At least he’s not Joe Biden.

4. Polls of non-Biden candidates are even shakier.

That said, the debate’s effects remain uncertain. The first debate in previous presidential cycles has shifted the polls, albeit modestly. And 2024 — like 2016 and 2020 — looks set to be a contest decided by fewer than 100,000 voters in a handful of states. If Biden loses the support of even a small portion of his base, his reelection chances will be at unusually high risk.

But those who suggest that other candidates would necessarily perform better don’t have much evidence to base that on. After all, this entire article has been about evaluating the potential effects on the established Trump-Biden contest, with the acknowledgment that those effects are difficult to predict. It’s harder to make assumptions about a contest between Trump and Vice President Harris, for example, because we don’t know how voters would react if she were at the top of the list. Add in other random, relatively unknown names and the ratings are even less reliable. For example, if you are very confident that you know Michigan Governor Gretchen Whitmer (D) would do better than Biden, you are operating on feelings, not data.

There’s another point worth making here: Questions about Biden’s electoral performance are different from questions about his fitness for office. To say that there’s little evidence that Biden would do significantly worse than other Democrats in a race against Donald Trump (which is true) is not to say that there’s no evidence that such a campaign or a second term would be any more fraught.

Instead, it says that what polls can tell us is limited and arrives more slowly than we would often like. We allow ourselves to be subjected to the whims and whims of people who insist they know what would happen even when they clearly cannot.