It’s still early in the primary season, but the whiff of possible election fraud is already in the air.
That’s because Donald J. Trump has underperformed the polls in each of his first three campaigns.
In Iowa, the final polling average of 538 showed Mr. Trump leading Nikki Haley by 34 points and 53 percent. In the end, he beat them by 32 points with 51 percent. (Ron DeSantis came in second.)
In New Hampshire, he led by 18 points and 54 percent. In the end he won by 11 points and 54 percent.
In South Carolina, Mr. Trump led by 28 points and 62 percent. In the end he won by 20 points and 60 percent.
In the scheme of the pre-election polls, these are not particularly big errors. In fact, they are more accurate than average.
But since Mr. Trump performed well against President Biden in early general election polls, even a modest underperformance by Trump in the polls is worth paying attention to.
So what’s going on? We can’t make a definitive statement based on the data we have, but three theories are worth considering.
One of them, described below, seems particularly plausible and is consistent with something we’ve written about before: anti-Trump voters are highly motivated to see out this cycle. That wouldn’t mean the polls would be wrong in November, but it would still be good news for Democrats.
Theory #1: Undecided voters
A simple explanation is that undecided voters ultimately supported Ms. Haley, the former governor of South Carolina.
That’s plausible. Mr Trump is a well-known candidate – even a de facto incumbent. If you’re a Republican and don’t know whether you support Mr. Trump at this point, you’re probably just not particularly fond of the former president. It’s easy to imagine how you might end up supporting his challenger.
It is also a theory supported by the survey patterns. Aside from Mr. DeSantis dropping out of the race, which caused that group of voters to shift toward Mr. Trump, support for Mr. Trump was stagnant in early states over the month leading up to this election. Over the same period, Ms. Haley has tended to post gains — gains that can most easily be attributed to undecided voters uniting behind her.
This was even true in South Carolina, where she narrowed the gap somewhat in the last round of polls.
Although this theory could easily be part of the story, it is not a complete explanation. In addition to having a narrower margin of victory compared to pre-election polls, Trump has tended to underperform his pre-election vote share, not simply because undecided voters are uniting behind Ms. Haley.
Theory #2: The electorate
Another possibility is that the polls simply misrepresented the composition of the electorate. In this theory, pollsters did a good job of measuring the people they wanted to measure, but they measured the wrong electorate. In particular, they did not include enough of the Democratic-leaning voters who supported Ms. Haley.
It’s impossible to prove, but I think this is probably an important factor. It is always relatively difficult to predict the makeup of the electorate in a presidential primary, but the large number of Democratic-leaning voters motivated to defeat Mr. Trump is particularly challenging this cycle. For the first time since 2012, there is no competitive Democratic presidential primary that could poach Democratic-leaning independents, and the Republican runner-up is a relatively moderate candidate who should be palatable to many Democratic-leaning voters.
We don’t yet have turnout data on how many Democratic-leaning voters actually turned out in these primaries, but there’s good reason to believe that’s part of what’s happening.
For many pollsters, the problem is entrenched from the start: They don’t even survey previous Democratic primary voters. Take, for example, the methodology of a Monmouth/Washington Post poll – one of the few polls whose methodology was disclosed in sufficient detail to allow accurate analysis:
The Monmouth University-Washington Post poll was conducted January 26-30, 2024, among a probability-based sample of 1,045 South Carolina voters who have voted in at least one Republican primary since 2016 or who have not re-registered since the 2020 election coordinated with an area code.
The decision to survey previous Republican primary voters is understandable – it makes the survey much cheaper and focuses on respondents most likely to vote – but it obviously leaves out all previous Democratic voters who did not vote took part in the Republican primary and are now deciding to do so.
How big is the problem for pollsters? It could be a big one. The pre-election voter turnout estimates we used for our Election Night Live model – you may know it simply as “The Needle” – assumed that 8 percent of Republican primary voters would be made up of former Democratic primary voters who had never voted for a Republican before. Primarily those who would not participate in the Monmouth/Washington Post poll. This group appears to have supported Ms. Haley.
That may seem like a lot for Democrats, but the final results suggest it may actually have been too low. In fact, the same turnout estimates clearly underestimated turnout in Democratic-leaning areas compared to Republican-leaning areas, suggesting that turnout among Democratic-leaning voters was even greater than expected.
The same goes for our turnout forecasts in New Hampshire last month: turnout in Democratic areas was a lot better than we expected. And realistically, the same challenge could continue to dog pollsters as long as primaries remain competitive, at least in open and semi-open primary states like South Carolina and New Hampshire.
Warning: Tuesday’s Michigan primary is also an open primary, although the campaign to vote “non-committal” to protest the war in Gaza may give Democrats a good reason to vote in their own primary.
There’s not much pollsters can do about this turnout problem. Many pollsters don’t have the money to survey the entire electorate for a low-turnout primary. Even if they poll everyone, they still have to conclude that these Democrats are likely to vote in a Republican primary, and I’m not sure that’s that easy to determine. How many of those voters, when asked by a pollster, would actually say something like, “I will almost certainly vote in the Republican primary”?
This is an unusual decision for Democratic-leaning voters, but one that many appear to be making.
If you are a Democrat hoping that the polls are underestimating Mr. Biden in the general election, your best case scenario is that the polls are wrong because there is a hidden Biden vote or at least a hidden anti-Trump vote.
According to this theory, the polls were good at modeling the electorate as undecided voters split between candidates, but anti-Trump voters simply didn’t participate in polls as frequently as pro-Trump voters. If that theory were true, then the general election polls could be underestimating Mr. Biden just as much as they underestimated Ms. Haley.
There is no good way to prove (or disprove) this theory. Typically, non-response bias theories gain credibility through an exclusionary diagnosis: once other explanations are ruled out, the possibility remains that the data exhibit unobserved bias. This is mainly because there is usually no clear evidence for non-response theories, which is also the case here.
The lack of evidence of non-response bias does not refute this. Far from it. But in this case, the turnout and undecided voter theories are credible enough that there is no reason to assume non-response bias either.
And realistically, neither the undecided voter nor turnout theories would have much influence on general election polls. There is no reason to believe that voters undecided between Mr. Trump and Mr. Biden will turn to Mr. Biden, at least not for the same reason that undecided Republicans might turn to newcomer Ms. Haley. The unusual turnout challenge for pollsters representing Democratic-leaning voters in open and semi-open Republican primaries also has no analogue to the general election.
There’s a reason anti-Trump turnout might be relevant to general election polls: It’s consistent with other data that shows Mr. Biden leading among the most engaged voters. This could result in a slight turnout advantage even in a general election. It may also mean that current polls of all registered voters slightly underestimate Mr. Biden compared to the narrower group of actual voters.
That wouldn’t mean today’s polls significantly underestimate Mr. Biden, but it could make all the difference in a close election.
Source link
2024-02-26 18:50:25
www.nytimes.com