Why you shouldn't sweat the 2024 polls — yet

Trump and Biden are basically tied. Yikes! But it's not time to panic.

Angela Weiss and Mandel Ngan/AFP via Getty

This is a special Saturday edition of Public Notice. Not a paid subscriber yet but want to read all of this post? Click the button below for sign up for a free trial.

This newsletter is entirely supported by readers, so please consider doing your part to make our work sustainable for the long haul.

---

It’s been a challenging nine months for Donald Trump. Last November, he led his party to a historically disappointing midterm showing — remember Dr. Oz and Herschel Walker? — and since then he’s been indicted four times and now faces a total of 91 felony charges.

President Biden, meanwhile, has been busy overseeing record job growth and an economy that’s proven to be remarkably resilient in comparison to others still struggling to pull out of the covid crash. America is clearly in much better shape than it was in January 2021. In a sane society it’d be hard to imagine the president losing to a candidate as flawed, unpopular, and legally troubled as the former guy.

---

RELATED FROM PN: The many reasons prominent Democrats won't primary Biden

---

Alas, America is a bit cuckoo. That’s borne out by a New York Times/Siena poll published earlier this month showing Biden and Trump heading into their likely 2024 rematch tied among registered voters at 43 percent. And that’s not an outlier. A recent CNN analysis found that “Trump has been ahead of Biden in more polls this year than during the entirety of the 2020 campaign.” Gulp.

The possibility of Trump returning to the White House and having the opportunity to finish off American democracy once and for all is obviously a nightmare, and it can’t be dismissed. But how worried should we really be about the polls this far out from November 5, 2024? To get to the bottom of that question, Public Notice contributor Thor Benson connected with a political data expert who was a voice in the wilderness during the run up to the 2022 midterms — Tom Bonier, CEO of TargetSmart.

While pundits last year predicted a red wave that would give Republicans big majorities in both chambers, Bonier was accurately pointing out that the numbers didn’t support Democratic doom and gloom scenarios, in part because pollsters weren’t accounting for how much Dobbs changed everything. The below tweet from June 2022 looks very smart in hindsight.

Source: the platform formerly known as Twitter

In short, Bonier knows what he’s talking about. And he doesn’t think you should lose sleep over the polls showing Biden and Trump tied.

“It’s been a grueling political climate in this country since Trump entered the scene, and there’s a lot of fatigue,” Bonier told us. “It’s not surprising that people look at a [2020] rematch and say they don’t want that. I do think that’s more about Trump than Biden, but those polls don’t capture the true dynamic we’ll see as we get closer to the election.” 

Bonier also shared his insights about the lessons of 2016, why polls overestimated GOP strength last year, and much more. A transcript of the discussion, lightly edited for clarity, follows.

Thor Benson

You were one of the few people who did not predict a red wave in 2022. Why did so many pundits get it wrong?

Tom Bonier

People were over-calibrating to past precedent, and historically the party that’s in the White House loses a whole lot of seats in midterms. The problem with that is we only have a finite number of midterm elections to draw from, and the sample is even smaller when you limit it to modern ones.

Comparing things to past precedent is important and relevant, but the fact is there wasn’t a great parallel for that election. We had just had the Dobbs decision, which was a huge deal. So it was an over-calibration to what people generally refer to as the “fundamentals,” which is past precedent, presidential popularity, economic indicators, inflation. All of those certainly pointed to Democrats losing.

Thor Benson

There’s been a lot of gnashing of teeth over the accuracy of polls over the last few election cycles. How much has polling improved, if at all?

Tom Bonier

2016 was probably the most stark example of the polling getting it wrong. It was an example of two elements that are still problematic today. One is systemic polling bias, and the other is how polls are analyzed.

When I say systemic bias, I mean the polls in 2016 were systematically overestimating Democratic support. From an analytical perspective, there was a layered bias there too. Because Clinton was winning so consistently in battleground states and in the national head-to-heads, it seemed impossible that she could lose. That said, credit to the polling industry for making changes to try to eliminate bias. The big one that everyone talks about is weighing more on educational attainment. That became a more important correlate with partisanship. 

Thor Benson

So things have improved?

Tom Bonier

Changes did lead to better results in 2018 and beyond into 2020. There were certainly improvements, but when you look at 2022, the conventional wisdom among analysts and pollsters and data journalists is that the polling was quite good. I’m not sure I share that view. And the challenge we have now looking ahead to 2024 is that there’s such a larger band of uncertainty.

2022 was the first election in some time where the systemic bias was more in favor of Republicans. I think that’s one reason why the analysis of the polling in 2022 led so many people to the conclusion that it was going to be a red wave election. They were looking at the surveys that showed a generally close election in the battlegrounds and in the Senate races and in the House, but then there was this general assumption that polls would overstate Democratic support. 

Because most people’s priors heading into that election was that Republicans were going to win, those things seemed to line up. Going into 2024, we know that bias in the polls is real. But is it going to be the kind of bias we saw in 2016 and 2020 where the polls overstated Democratic support? Or is it going to be the kind of bias we saw in 2022 where it overstated Republican support? That’s one challenge which will lend itself to a larger degree of uncertainty. 

Another challenge is that we play this game after each election of assessing which polls were accurate. The sense was that the New York Times/Siena poll was much more accurate in 2022. In other years other polls were more accurate. When you look at the 538 pollster ratings, you’ll see some surprising results. Some pollsters that performed very badly in 2022 have A ratings or B ratings. When you’re analyzing individual polls and the past track record of pollsters, we’ve learned that one or two election cycles isn’t necessarily indicative of their accuracy in a future election. 

Thor Benson

We’ve seen some polls recently where a significant number of Republicans have said they might not vote for Trump if he’s indicted or convicted. That seems surprising. Do you think we can trust those polls?

Tom Bonier

I don’t. I’m not a big fan of that type of question. People are poor predictors of their own behavior. With questions like that, there’s this social desirability element. There’s this notion that perhaps there’s a way someone thinks they should answer or how they would like to think they would react. In the end, until you have that lived experience of seeing a former president and presidential candidate indicted and potentially convicted, no one’s truly able to predict the impact that will have on their voting decision.

Presumably, if you see someone getting indicted and potentially convicted and that makes you more likely to support them, you were already supporting them. I’m not sure it’s something that’ll be that helpful other than potentially mobilizing more of his supporters that may have some level of fatigue. This might, in some ways, revive their enthusiasm. There are some people who held their noses and voted for him in the past, and this might be their tipping point. In general, I wouldn’t pay much attention to those types of polling questions. 

Thor Benson

There are also polls saying people don’t want a Trump vs. Biden rematch. They’re not very popular as candidates. How much do you think we should pay attention to those polls?

Tom Bonier

Similarly, I wouldn’t pay a whole lot of attention to that. I would even extend that to the head-to-head polling we’re seeing now for a Trump/Biden rematch. There’s a lot of attention paid to those polls, especially when there’s one that shows that Trump is close. The fact is that election is still so far away. The reality is that until people actually experience the campaign and really think about going into the voting booth and choosing one or the other, it’s impossible to know what’s going to happen. I’d say ignore those polls, but I’d be a hypocrite to say that. I certainly look at them. 

We just shouldn’t put too much weight on any of those surveys this far out. It’s been a grueling political climate in this country since Trump entered the scene, and there’s a lot of fatigue. When people think of a rematch of the 2020 election, I don’t think many people felt good about that. It’s not surprising that people look at a rematch and say they don’t want that. I do think that’s more about Trump than Biden, but those polls don’t capture the true dynamic we’ll see as we get closer to the election. 

---

That’s it for today

We’ll be back with more Monday. Have a great weekend, and thanks as always for your support.

---