Julia, I'm suspicious that the reason Trump's lead in primary polling has been somewhat exaggerated is that pollsters are not necessarily capturing some of the independents who are showing up to vote in a GOP primary. As Nate Cohn recently pointed out in The New York Times, some pollsters have filtered for likely GOP primary voters by mostly or only looking at voters who've voted in Republican primaries. But a fair number of independents, including Democratic-leaning ones, wouldn't necessarily fall in that category. But with Biden a lock to win the Democratic nomination — barring something happening outside of the voting booth — Democratic-leaning voters have only one race to get involved in. Although we know there aren't that many "cross-over" voters who participate in the opposing party's primary, some high-propensity Democratic voters and some Democratic-leaning independents have probably voted in states that allow non-Republicans to cast ballots in the GOP primary.

—Geoffrey Skelley, 538

In 2016, people who said they were undecided in polls were more likely to vote for Trump as "professor polls" explains in this great 538 video I show my students. But in 2024, Trump supporters aren't so shy. It's the rest of the GOP that might be more bashful.

—Meredith Conroy, 538 contributor

I actually don't think polls are overestimating Trump by much, more that they're underestimating Haley. And I think it really comes down to undecided voters. Trump is a known quantity for pretty much everyone in America. So anyone who isn't sure if they're going to vote for him isn't unaware of his issues, but probably has some doubts about him. So my theory is that undecided voters skew, if not anti-Trump, at least Trump-skeptical and are disproportionately choosing Haley. You can see an inkling of that in the preliminary exit poll data: In Virginia, for example, voters that decided their vote choice earlier than this year went overwhelmingly for Trump (78 to 21 percent), but voters that made their decision later chose Haley (54 to 42 percent).

—Mary Radcliffe, 538

Polling nowadays is incredibly hard. Seeing response rates of under 2 percent is all too common. So even very slight differences in how fired up people are to take polls can translate into meaningful differences between polls and results. Primary polling is especially hard because you can't use partisan identification to benchmark how reasonable the sample looks. That all said, in New Hampshire and in South Carolina, I wonder if Haley may have benefited from the lack of a competitive Democratic primary, and the fact that some who might have otherwise voted in a Democratic primary were able to vote for Haley.

—Dan Hopkins, 538 contributor