Would young Americans really vote for Trump today?

Why I’m skeptical of some puzzling polls

Crosstab averages from split-ticket.org

Originally I was not going to soft-publish a blog post today. My wife and I both have somewhat serious cases of covid and I’d rather try to sleep than push out a few hundred to a thousand words about polls or the news. But, after stomaching a real meal and taking an ungodly dose of cold medicine, I have some energy. I also got name-dropped in an article today and wanted to respond.

Longtime columnist for New York Magazine Jonathan Chait writes in his article “Are Young People Actually Progressive?” that I have played a part in peddling some bad assumptions that are putting the Democratic party “in peril”:

For the past two decades, young people were widely assumed to have an ironclad loyalty to the Democratic Party. Democrats believed this, as did Republicans and journalists. So deeply ingrained was this belief that when polls began to show President Bidenlosing to Donald Trump in the 2024 race, critics sometimes disregarded them on the grounds that these surveys showed young people favoring Trump — a result nobody could accept.

When a Washington Post–ABC News poll in September found a ten-point Trump lead, for example, analyst G. Elliott Morris dismissed it as an outlier in part because it showed voters between the ages of 18 and 29 favoring Trump over Biden 48-to-41 percent. Morris noted that an ABC News exit poll from the 2020 election showed Biden winning that group by 24 points, which would mean a 31-point swing toward Trump in the past few years — “more than twice as much as the overall population shifted.”

Morris conceded it was possible “if young people and people of color are dissatisfied with the progress Biden has made or upset that he is running for reelection, they could be withholding their support for him in polls as a way to register their displeasure with the situation. Ultimately, however, they may vote for him anyway.” In other words, young voters might say they are voting for Trump because they are upset Biden isn’t liberal enough. But they couldn’t really mean it. Right?

It seems, alas, that they do. A New York Times poll published in December is the latest to find Trump leading Biden among young voters. The unthinkable has become, at least for the time being, undeniable.

(For now I’m setting aside the problematic error here that one poll makes something “undeniable,” but I’ve bolded it to come back to this later.)

Chait then goes on to write about the mistaken theories of political polarization pushed by demographer Ruy Teixeira and journalist John B. Judis in their decades-old book The Emerging Democratic Majority. To the columnist, these theories caused party leaders and activists alike to pour untold resources into mobilization-focused tactics to get young and non-white voters to the polls. But as racial polarization has peaked and education polarization has taken its place, these groups, the argument goes, are not a reliable base for Democrats after all. He goes on to say that high-turnout elections have not produced significantly better results for Democrats recently.

To be fair to me, I’m not sure why I’m called out or have anything to do with this piece. I did not write The Emerging Democratic Majority and do not set Democratic Party strategy agendas. If you actually read my article from late September you’ll see that all we were doing was trying to explain our skepticism about a subgroup crosstab that — at the time — no other poll had found. Voting behavior typically doesn’t change by 30 points for subgroups between two election cycles, and when it does, swings are usually found in other groups too. I stand by that argument regardless of the single Times poll Chait cites to make his argument.

Partisan non-response and “expressive responding”

In fact, what Chait is doing here betrays a misunderstanding of the current critique of the polls in regards to their samples of young people. In my article I suggest that young people may be misrepresented in vote intention for two reasons. The first is partisan non-response; Since most polls do not ensure subgroups are balanced by party, it’s possible that young 2020 Biden supporters in these ABC News/Washington Post and New York Times polls simply aren’t answering the phone, or that the 2020 Biden voters they are getting are especially anti-Biden now in ways that are unrepresentative. In the case of ABC we can’t know for sure, because we don’t know the characteristics of the people not answering the phone. The Times maintains its subgroups are politically representative, though as far as I know has not released anything about young non-voters (which is where I found the non-response in the ABC data).

The second possibility is that the young people responding to vote-intention polls are thinking of other, non-vote-related information when pollsters are calling them, and then layering responses that that information on top of some true underlying feeling. In the polling literature this sort of layering can take many forms, but the term that has been thrown around recently is “expressive responding.” Expressive responding dictates that people answering a poll may give you an attitude they don’t truly feel in order to make a point; Say, a Republican hesitant about Trump saying they view him favorable to go with the crowd, or a young person saying they’d vote for Trump because they’re mad at Biden on student loans or Gaza. This vibes with what we know about voter psychology and responding to polls.

Again there is no reason to expect that these sorts of biases would affect one poll and not another (of the same mode). This means that citing the NYT poll to take down the youth-quake naysayers is not the slam dunk Chait thinks it is. These two data points are not proof that young people aren’t progressive anymore, as they could be biased by the same underlying factors in their data-generating processes. We would have to very uncritically accept the polls at face value to buy that — and recent elections have shown how unwise that would be.

Wait, do we really think young people aren’t progressive?

The idea that more voters under the age of 30 would cast ballots for Donald Trump than Joe Biden starts to fall apart even more when you think about what it would mean for their ideology. Given the headline of his piece, Chait seems to be one or two polls away from buying the idea that young people aren’t even “progressive.” But thanks to work by my colleagues at 538 we know just how monumental that shift would be.

Young voters under 30, according to 538’s Monica Potts and Holly Fuong, are increasingly left-leaning on climate change and the environment, abortion, and immigration. They write:

Sixty-four percent of young voters in the CES data said they supported a policy that "always allow a woman to obtain an abortion as a matter of choice" in 2020, compared with 56 percent of the general population. Those numbers increased to 70 and 59 percent in 2022, respectively.

and

In 2020, 78 percent of young voters supported granting legal status to all illegal immigrants who have held jobs, paid taxes and not been convicted of a felony, 10 points higher than the general electorate. They were also less likely to support increasing border patrols, withdrawing federal funds from police departments that don't report the immigration status of detainees, reducing legal immigration and increasing funding for border security and building a wall. The differences were even bigger in 2022.

Potts and Fuong also make the crucial point that Generation Z is achieving higher rates of education than priors generations. That would push them left, all else being equal, thanks to education polarization.

Just to reiterate, these numbers showing huge proportions of young Democrats are progressives are as of the 2022 midterms. To believe that Trump really is winning young voters, that young people aren’t progressives, would be to believe that mass opinion among the group has changed by 30 or 40 points in the span of a year — an already hard-to-believe shift, and that’s before considering these polls show no other groups are moving double digits to the right.

How to update your priors

So if we don’t really believe these polls, what are we to do with them? I personally think exercises like this blog post are helpful. When data don’t fit our priors it’s worth considering the way the data were generated, and where our models of the world might be breaking down. Polls are just models after all, and they come with all the caveats you would expect when someone tells you they can represent an entire country’s opinions with a sample of just 1,000 people (with a dismal response rate).

You definitely don’t want to simply uncritically accept the polls as truth,. But, equally bad, would be to discard the data altogether. In a Twitter/X post today responding to new crosstabs from a Suffolk University poll, Democratic strategist and commentator Simon Rosenberg called for such polls to be thrown out entirely:

Now this would be bad for several reasons. I explained why back in my September article on outlier polls:

That’s a dangerous precedent to set: Throwing out polls you don’t believe in is a recipe for “herding” — where pollsters release their polls only if they’re close to other surveys (or, in the worst cases, actually adjust their polls so they match others). This can decrease the accuracy of polling averages, which rely on seeing a diverse portrait of public opinion to find the right signal in the data. More philosophically, the entire point of scientific polling is to root your understanding of politics in some sort of objective scientific process. When that process yields uncertain results, you should question them and rework any errors, but it takes time and repeat observations to know when something is broken.

That’s why 538’s philosophy has long been to average outliers (warts and all) with all the other surveys we have of the same race. If you have enough data (as we do in high-profile elections like the 2024 presidential race), then random outliers will barely budge the bottom-line average.

Of course, the idea here is that these polls are not “random outliers” but are being pulled around by the same DGPs. And on that note I’ll link to this week’s YouGov/The Economist poll, which was conducted online (rather than over the phone, per ABC and NYT) and shows young people voting almost 20 points for Biden over Trump if the election were held today. These differences by mode are exactly what you’d expect if there was some sort of phone-based nonresponse or expressive responding by young people.

Just to make this very clear: This is not an argument to throw the baby out with the bath water. What you really want to do is update your prior beliefs about the world some amount based on the data. I tweeted out some formulas for this earlier this year: Let’s say your prior about young people is Biden +26, and the crosstabs of polls are currently showing Trump +2 on average. Maybe you think the sample sizes are too noisy or these other non-sampling factors are biasing the data, so you don’t put too much weight on them. In that case you’d end up around Biden +16 for young people in the 2024 election. Based on the exit polls from 2012, 2016, 2020 and 2022, I think that’s probably pessimistic — but the key here is that we start thinking about distributions of truth as revealed by the data rather than uncritically adopting or completely rejecting it.

Finally, note that it’s early, and most people aren’t thinking about the election at all right now. If polls show these same patterns in October 2024, I will update my priors a little more. But until then you can count me firmly in the camp that thinks young people are disproportionately progressive and polls are overestimating their true support for Trump.