Three theories about why President Trump's key results don't match expectations.

It's still early in the primary season, but the possibility of voting errors is already in the air.

That's because Donald J. Trump has trailed in the polls in each of his first three campaigns.

• In Iowa, the average vote count of the final 538 votes gave Trump a 34 point lead over Nikki Haley, giving him a 53% share. In the end, he beat her by 32 points with 51 percent. (Ron DeSantis took second place.)

• In New Hampshire, he led by 18 points at 54 percent. In the end, they won by 54 percent, or 11 points.

• In South Carolina, Trump led by 28 points with 62%. In the end, they won by 60 percent, or 20 points.

In the scheme of preliminary polls, these aren't particularly big outliers. In fact, they are more accurate than average.

But Mr. Trump has performed well against President Biden in early polls, and even if Mr. Trump is trailing slightly in the polls, he deserves some attention.

what happened? Although we cannot say anything conclusive based on the data available to us, three theories are worth considering.

One of them, explained at the bottom, seems particularly plausible and is consistent with what I wrote earlier. So anti-Trump voters are very motivated to solve this cycle. That doesn't mean the November polls are wrong, but it's still good news for Democrats.

Theory #1: Undecided voters

One simple explanation is that undecided voters ultimately supported Haley, the former South Carolina governor.

This is plausible. Trump is both a well-known candidate and the de facto incumbent. If you're a Republican and currently unsure whether you support Trump, you probably just don't have any particular fondness for the former president. It's easy to see how you'll end up supporting his challenger.

This is also a theory that has some support in polling patterns. Trump's approval ratings in the early states remained flat for about a month before the election, except for DeSantis' withdrawal from the race and the shift of voters to Trump. During the same period, Mr. Haley tended to make a profit. That gain is most easily attributed to undecided voters rallying behind her.

The same was true in South Carolina, where he narrowed his lead slightly in the final vote.

But while this theory could easily be part of the story, it's not the complete explanation.In addition to a lower win rate compared to pre-election polls, Trump was more likely to run. behind Her pre-election vote share was not simply due to undecided voters rallying behind Haley.

Theory #2: Voters.

Another possibility is that the polls simply got the composition of the electorate wrong. According to this theory, pollsters were doing a good job of measuring the people they were trying to measure, but they were measuring the wrong voters. In particular, it did not include enough Democratic-leaning voters who had come to support Haley.

It's impossible to prove, but I think this is probably a major factor. Predicting the composition of the electorate in presidential primaries is always relatively difficult, but the large number of Democratic-leaning voters who are motivated to defeat Trump poses a particular challenge this season. For the first time since 2012, there is no competitive Democratic presidential primary that separates Democratic-leaning independents, and the Republican runner-up is a relatively moderate candidate that may appeal to many Democratic-leaning voters. unknown.

There is no turnout data yet on how many Democratic-leaning voters actually participated in these primaries, but there is good reason to believe this is part of what is happening now.

For many polling organizations, this issue is built in from the beginning, and they don't even interview former Democratic primary voters. For example, consider the methodology of the Monmouth/Washington Post poll. This is one of the few polls that discloses a methodology detailed enough for close analysis.

of Monmouth University-Washington Post Poll The survey will be conducted from January 26 to 30, 2024, for those who have voted in at least one Republican primary since 2016 or who have registered since the 2020 election but have not voted in a primary. It was conducted on a probability-based sample of 1,045 Carolina voters.

While the decision to survey former Republican primary voters is understandable, as it would be much cheaper to survey and would narrow down the respondents who are most likely to vote, those who have not voted in the Republican primary , would clearly miss out on former Democratic voters who chose to vote this time. So.

How much of a problem is this for pollsters? It could be a big deal. The pre-election turnout estimates we used for our election night live model (sometimes known simply as the Needle) show that 8% of Republican primary voters have ever voted Republican. It was assumed that it would be comprised of former Democratic primary voters. People who are not eligible to participate in the primary election, Monmouth/Washington Post poll. It seems likely that the group supported Mr. Haley.

While this may seem like a high number of Democrats, it suggests that the final result may actually have been too low. In fact, these same pre-election turnout estimates clearly underestimated turnout in Democratic-leaning areas relative to Republican-leaning areas, suggesting that turnout among Democratic-leaning voters was even more active than predicted. It suggests that.

The same was true for turnout projections in New Hampshire last month. Turnout in Democratic districts was much better than we expected. And realistically, at least in open or semi-open states like South Carolina and New Hampshire, the same challenges will continue to haunt pollsters as long as primaries remain competitive. There is sex.

Warning: Tuesday's Michigan primary is also an open primary, but a campaign to vote for “irresponsible” people protesting the war in Gaza may give Democrats enough reason to vote in their own primary.

There's not much pollsters can do about this turnout issue. Many polling organizations don't have the funds to survey the entire electorate in primaries with low turnout. Even if we surveyed everyone, we would have to conclude that these Democrats were more likely to vote in the Republican primary, but I'm not sure that's that easy to determine. If asked by a pollster, how many of these voters would really say something like, “I'm almost certain to vote in the Republican primary?”

This is an unusual decision for Democratic-leaning voters, but it appears many voters are making the decision.

Theory #3: Hidden Biden votes?

If you're a Democrat who's hoping the polls are underestimating Biden in the general election, you're saying the polls are wrong because there are hidden Biden votes, or at least hidden anti-Trump votes. That's the best case scenario.

According to this theory, polls do a good job of modeling voters, with undecided voters split between candidates, but anti-Trump voters responding to surveys as much as pro-Trump voters. It was unlikely. If this theory is true, the general election polls may be underestimating Biden just as much as they are underestimating Haley.

There is no good way to prove (or disprove) this theory. Nonresponse bias theories typically gain credibility through exclusionary diagnoses. When other explanations are ruled out, it leaves open the possibility of unobserved bias in the data. The main reason is that nonresponse theories usually lack clear evidence to support them, and the same is true here.

Just because there is no evidence of nonresponse bias does not disprove it. Far from it. But in this case, the turnout and undecided voter theories are sufficiently reliable that there is no reason to assume nonresponse bias either.

And realistically, neither undecided voters nor turnout theory will have much of an impact on general election voting. There is no reason to expect voters who are undecided between Trump and Biden to lean toward Biden. At the very least, there is a possibility that Republicans who are still undecided may lean toward the newcomer Haley. The unusual turnout challenges posed to pollsters by Democratic-leaning voters in open and semi-open Republican primaries also bear no similarities to the general election.

There is one reason why anti-Trump turnout may be relevant to general election polls. That's consistent with other data showing Biden's lead among the most engaged voters. This could give a slight advantage to voter turnout even in general elections. It also could mean that current polls of all registered voters slightly underestimate Mr. Biden compared to a narrower group of actual voters.

This doesn't mean today's polls are significantly underestimating Biden, but it could make a difference in a close election.