We Did an Experiment to See How Much Democracy and Abortion Matter to Voters

Do abortion and democracy matter to voters?

If you look at the results of New York Times/Siena College polling, the answer often seems to be “not really.”

Around 40 percent of voters agreed that Donald J. Trump was “bad” for democracy in our latest poll. Only around a quarter said that issues like democracy and abortion were more important to their vote than the economy.

But in election after election, the final vote tallies seem to tell a very different story. Last fall, Democrats excelled when abortion and democracy were at stake, even though our pre-election polls offered little indication that these issues were driving voters. It raises the possibility that the usual poll questions simply failed to reveal the importance of abortion, democracy and perhaps other issues as well.

With that in mind, we tried an experiment in our latest Times/Siena poll. We looked at the persuadable voters — those who were undecided or who said they were open to supporting the other candidate — and split them into two groups. We gave each group a set of two hypothetical Republican candidates based on views on abortion and democracy.

While only an experiment, the findings suggest that democracy has the potential to be an extremely important factor in people’s voting — even among voters who say it’s not important to them at all.

Here’s the democracy matchup:

• Hypothetical B: Would you be more likely to support a Democratic candidate who says Donald Trump is a unique threat to democracy, or a Republican candidate who says we should move on from the 2020 election?

If democracy didn’t matter to voters, these two hypotheticals might not yield very different results.

But the results were vastly different.

In the experiment, the hypothetical anti-Trump Democrat in Group A led by three percentage points against a Republican who tried to overturn the election, compared with Mr. Trump’s five-point lead over President Biden across the full sample in the poll.

On the other hand, the “move on from 2020” Republican (Group B) led by 15 points against the same anti-Trump Democrat.

Now, perhaps this hypothetical exaggerates the effect of the issue. In the real world, voters will consider a lot more than democracy — and this hypothetical certainly draws the respondent to focus on democracy. Indeed, we estimated that MAGA candidates fared five points worse than non-MAGA candidates last fall, not 18 points (though perhaps the “move on from 2020” candidate is more moderate than the average non-MAGA Republican, and perhaps the “tried to overturn the election” candidate is more extreme than the typical MAGA election denier).

But the experiment nonetheless shows that a lot of voters might respond very differently to Republican election subverters, even though the same voters gave answers that could be interpreted to mean the issue wasn’t necessarily important to them.

Strangely, a majority of voters who flipped against the election subverter told us that they didn’t think Mr. Trump was bad for democracy. By a three-to-one margin, they said the economy was more important than democracy and abortion.

This is not what I expected. I thought we would find that there was a portion of voters who consistently cared about democracy, but that they simply weren’t a majority of registered voters. Instead, we have a puzzling group of respondents who don’t seem to care that much about democracy based on our typical questions, but then act as if they do in a hypothetical matchup.

This is difficult to explain, but given the real world election results it’s reasonable to think that the problem lies with our typical questions, not our experiment. After all, the hypothetical candidate matchup comes closest to presenting voters with the actual choice they face at the ballot box. And so it doesn’t surprise me that it yields the answer that most closely approximates real world electoral outcomes.

In contrast, our version of this experiment on abortion didn’t find nearly as significant movement. That’s partly because the difference between the two Republicans isn’t as extreme, but here it is:

• Hypothetical A: Would you be more likely to support a Democratic candidate who supports a federal law ensuring access to abortion nationwide or a Republican candidate who supports a federal ban on abortions after 15 weeks of pregnancy?

• Hypothetical B: Would you be more likely to support a Democratic candidate who supports a federal law ensuring access to abortion nationwide or a Republican candidate who says abortion should be left to the states?

The Republican presidential candidate who supported a 15-week ban (Group A) had a one-point lead. That’s worse than Mr. Trump’s five-point lead in the poll, perhaps confirming that the issue isn’t especially favorable to Republicans — though it’s not exactly fantastic for Democrats.

Meanwhile, the Republican who promised to leave abortion to the states (Group B) had a four-point lead — roughly the same as Mr. Trump’s standing against Mr. Biden, suggesting Republicans can mostly defuse the issue by saying they’ll leave it to the states.

In retrospect, I wish we had chosen a more extreme conservative position on abortion for this experiment, just to see whether we would find a bigger difference.

But the narrower gap between the hypothetical candidates reflects how the aspiring Republican presidential candidates are handling the issue. And the relatively narrow difference on abortion lines up with the election results. While it’s clear voters are highly supportive of abortion rights — including in these polls — it can be hard to find clear signs of voters punishing anti-abortion Republicans.

Ohio is a good example. By a 13-point margin, it recently voted to protect access to abortion. Clearly, voters in the state support abortion rights. Yet last fall, it re-elected the Republican governor who signed the abortion ban, Mike DeWine, by 25 points. In contrast, the MAGA candidates were dealt a plain electoral penalty — including in Ohio. A similar story played out in other states with abortion bans, like Texas or Georgia.

None of this means that abortion didn’t or doesn’t help Democrats. If Roe hadn’t been overturned, the national political conversation might have focused more on issues helping Republicans, like the economy or crime. This probably helped Democrats almost everywhere, regardless of the Republican candidate’s view on the issue.

What it does mean is that anti-abortion Republicans don’t seem to suffer the same clear electoral penalty that the MAGA election deniers or subverters face. Voters may not say democracy is the most important issue, but there’s strong evidence that they’re repelled by these candidates. The Biden campaign will surely try to remind voters that Mr. Trump is one of them.