Do any Trump voters regret their choice?

It's a hard question to answer. Our poll suggests many do.

Dear readers,

Here's a question that has been raised to me a few times over the last month:

Would Donald Trump win the 2024 election if it were held again today?

I know this seems like a silly question (it will never be 2024 again and Trump is president), but the answer is potentially very revealing about both the state of public opinion on the Trump administration and the potential electoral consequences for the Republican Party in next year's midterms.

Regret is also a powerful collective political-psychological tool. We would collectively feel a lot differently about our country, and about each other, if 100% of Trump voters are happy with their decision versus if only 50% are happy. And there is an accountability mechanism here too; Trump himself would also presumably feel a lot different if half of his voters were regretful, versus if he had a 100% approval rating. In the latter scenario, he is emboldened, the strong leader enacting a popular mandate for the country — and, importantly, his people. In the former, eh, maybe he doesn't have as much leverage in Congress (or something else meaningful).

So then, what does the data say?

On one hand, the president's low approval rating, especially on the economy, suggests he probably wouldn't win. The 2024 election was close in the swing states, and Trump's job approval today is lower than it was for much of the 2020 election, when he lost by 4.5 points in the popular vote. And given what voters know now about his likelihood to lower prices (perhaps the key issue of the 2024 campaign), he may have lost his biggest advantage.

But on the other hand, some polls show very few Trump voters actually feel regret over their vote last time. A particularly viral recent cable news clip is illustrative: CNN's Harry Enten recently used a poll conducted by the University of Massachusetts-Amherst in April to show that "very few” Trump voters “regret what they did back in 2024." In that poll, when asked directly if they regretted their vote, just 4% of self-described Trump voters say they would have voted for someone else or not at all, with 93% saying they feel confident about their vote.

Based on those numbers, Enten argues:

“So if there was a repeat, if folks got to be able to redo their vote that they had back in 2024, would the result be any different? I doubt it would be. I doubt it would be or it would still be extremely close. The bottom line is for all this talk of Trump voters regretting their vote, in the numbers, it really just doesn’t show up.” [My emphasis added]

Well that's technically an accurate reading of the poll, but, the thing is... I don't think these numbers actually evidence his point about a 2024 re-do, and it ignores a crucial third category of voters that would make the difference if the election were re-held today: non-voters. So let's take a look at this, in today's Chart of the Week.

---

---

Don't ask a man how much he makes, a woman how much she weighs, or if an Italian likes pineapple on pizza

Broadly, there are two problems with the UMass-Amherst poll being used in this way. And I’m not singling out Enten; USA Today also had a piece up about low levels of Trump voter regret. It was all over my social media and in various Substacks, etc.

The first problem is that people are probably very unwilling to tell a stranger that they regret an action they took. Poll results often exhibit signs of "social desirability bias," where respondents prefer to give desirable responses instead of expressing their true feelings. One example of this is men exaggerating their height, and pretty much everyone exaggerates their income. People want to impress strangers, and they also want to save face with them.

Accordingly, it seems relatively unlikely to me that a Trump voter who feels some regret about their choice would want to admit that to some random person on the internet. And that's especially true, in my estimation, when the stakes of the decision are so high.

So from first principles, I don't think the poll is truly measuring what we want it to measure. And given the magnitude of potential effects from social desirability bias, it’s worth discounting it just based on these grounds.

Trump voters are less loyal, and non-voters tilt the scales to Harris

The second problem with this story from Enten and others is that this calculation excludes 2024 non-voters. If the 2024 election were held again today, a lot of people who stayed home would have new information about the consequences of their decision, and many of them would behave differently.

If political events are an independent shock to vote choice, they are probably a shock to turnout, too.

Thanks to a question submitted by a loyal Strength In Numbers subscriber to this month's Strength In Numbers/Verasight survey, we have some actual data that attenuates biases from both problems (2024 past-vote and social desirability bias).

In a poll of 1,000 U.S. adults, I find that if you ask people simply how they would vote in a hypothetical election (not framed as regret), persistent voting for Trump is in fact significantly lower than persistent support for Kamala Harris. The percentage of Trump 2024 voters who say they'd vote for him again is just 86%. In comparison, Harris retains 92% of her prior voters. Those differences seem small, but in a close election, they absolutely add up.

At least according to their revealed preferences, Trump voters do seem to be somewhat to significantly regretful — and potentially even more regretful than Harris voters, though these percentages are within the margin of error of each other. The cross-over voters are also killer for Trump; 5% of his 2024 supporters say they'd vote for Harris if the 2024 election were held again today, compared to 1% for Trump.

But what people tend to ignore is the non-voters. A meaningful number of people who sat out 2024 now say they would participate, and they lean toward Harris over Trump by 14 points. While 2024 voters are split 43% to 43% for Trump and Harris in the hypothetical rematch, non-voters go for Harris 36% to 22%.

Non-voters represent about a third of U.S. adults, so that shift is large enough to tip the electoral balance away from Trump overall (as reported in our first release of Verasight data).

The “regret” that matters most may not be among Trump’s base, but among those who didn’t vote at all.

It's just one poll, but it's actually about the thing we care about

This was a fun article to write for two reasons. First, it's a fun example of the type of insights we can generate here at Strength In Numbers when we do our own data collection and in-depth survey analysis. And not only that, this question was recommended by a subscriber to the newsletter, so I feel a strong author-audience relationship via this COTW — and it’s a good advertisement for what readers get in return for their participation.

Thinking about the bigger picture, though, this is the type of analysis that I think sets Strength In Numbers apart from the competition. Nobody in independent data-driven political media is doing their own surveys, analyzing their own toplines and crosstabs, to answer questions like these for readers in a transparent and honest way. There’s value there.

This is also a reminder that simple narratives about polls can often mislead us. Yes, Trump voters tell us they don't regret their choice. But do they mean that? What do their revealed preferences say? What do they tell us about other questions that might have a closer link to things we care about that we can observe in the real world (eg, election outcomes)?

And, sure, this is just one poll. I'd love to see pollsters publish the table above using their data, so we can get a sense of whether this replicates across samples. These estimates come with margins of error, and more data would reduce that error, etc. There’s also the point that not all 2024 non-voters would actually turn out to vote in an election (they didn’t vote in the last one!), so the comparable I’ve created is not a perfect mapping onto reality. Well, yeah, it’s a hypothetical.

But to the extent that what we are after here at Strength In Numbers is a better approximation in general of "the ground truth” of the world, I think this gets a lot closer to what we care about than the existing work. In this case, despite the existing narrative, loyalty to Trump does seem particularly low. It is likely lower than loyalty for Harris.

You can think of many explanations for that, from moving policy too far to the right and causing a thermostatic shift to whatever pet explanation fits your priors. The precise explanation is not really the point here. The point here is that it pays to be careful and accurate when talking about public opinion — and in this case, new data points strongly against the president.

---