Who will have the momentum out of Iowa?
On a blisteringly cold Monday in Iowa, Republican caucusgoers handed former President Donald Trump a huge victory, the first step in a long — but increasingly uncompetitive — presidential primary race. Trump won 51 percent of the vote (right in line with the polls), sweeping all but one county: Johnson County, the home of Iowa City and the University of Iowa. Distantly trailing him were Florida Gov. Ron DeSantis at 21 percent, former U.N. Ambassador Nikki Haley at 19 percent and entrepreneur Vivek Ramaswamy at 8 percent.
Most of this was, to say the least, utterly predictable. But there was one mild surprise: DeSantis. His 21 percent of the vote was 5 percentage points higher than his final Iowa polling average. That is not a huge polling error by historical standards, but it nevertheless gave the governor enough ammo to claim momentum heading into the next few primary contests.
But what does DeSantis's slight overperformance mean for his odds of winning the primary? We have found that, once Iowa rolls around, candidates polling at 63 percent nationally (as Trump is doing) very, very rarely (like, 5 percent of the time at most) go on to lose their party's nomination. That puts Trump in a dominant position regardless of the results in Iowa. At the same time, Iowa is one of the rare campaign events that can shake up the remainder of the primary. Think about then-Sen. Barack Obama's rise after a surprise win in Iowa in 2008, or former Sen. Rick Santorum's surge after winning the caucuses in 2012. In both cases, strong Iowa results created a sense of momentum for the candidates going into subsequent nominating contests.
But because we're empiricists here at 538, I booted up our historical polling and election results database to quantify just how much momentum candidates receive in Iowa when they beat expectations. The answer actually surprised me somewhat; it's not often that we find results in polling and political analysis that line up so cleanly with a conclusion.
A good Iowa performance can spur national gains
The table below compares how much candidates over- or underperformed their polls in Iowa with how much they gained or lost in national polls over the subsequent month, based on retroactively calculated 538 polling averages from 1980 through 2020. The table includes only candidates who finished first or second in Iowa.
Beating expectations in Iowa can create national momentum
538 polling averages and final vote shares in Iowa for the top two finishers in the Iowa caucuses from 1980 to 2020, and the change in their 538 national polling averages over the subsequent month
Iowa | ||||||
---|---|---|---|---|---|---|
1980 | Bush | 1 | 12.1% | 31.6% | +19.5 | +22 |
2004 | Kerry | 1 | 23.7 | 37.6 | +14.0 | +33 |
1988 | Robertson | 2 | 11.5 | 24.6 | +13.1 | 0 |
2004 | Edwards | 2 | 18.9 | 31.8 | +12.9 | +10 |
1996 | Buchanan | 2 | 10.6 | 23.3 | +12.7 | +15 |
2012 | Santorum | 1 | 13.2 | 24.5 | +11.4 | +12 |
2008 | Obama | 1 | 26.4 | 37.6 | +11.1 | +15 |
1980 | Carter | 1 | 48.5 | 59.1 | +10.6 | +1 |
2020 | Buttigieg | 1 | 15.7 | 26.2 | +10.4 | +3 |
1984 | Hart | 2 | 6.3 | 16.5 | +10.2 | +32 |
1988 | Simon | 2 | 17.1 | 26.7 | +9.6 | -4 |
2000 | Gore | 1 | 57.2 | 63.4 | +6.2 | +10 |
2008 | Edwards | 2 | 23.7 | 29.8 | +6.0 | -2 |
1988 | Gephardt | 1 | 25.5 | 31.3 | +5.8 | +9 |
2016 | Sanders | 2 | 43.9 | 49.6 | +5.7 | +1 |
2008 | Huckabee | 1 | 28.8 | 34.4 | +5.6 | -4 |
2016 | Cruz | 1 | 22.7 | 27.6 | +5.0 | +1 |
2000 | Bradley | 2 | 30.6 | 34.9 | +4.3 | -10 |
2020 | Sanders | 2 | 22.2 | 26.1 | +3.9 | +6 |
2016 | Clinton | 1 | 47.3 | 49.8 | +2.5 | 0 |
2012 | Romney | 2 | 22.5 | 24.5 | +2.0 | +5 |
1996 | Dole | 1 | 25.6 | 26.3 | +0.7 | -7 |
2008 | Romney | 2 | 25.8 | 25.2 | -0.6 | +7 |
1988 | Dole | 1 | 38.3 | 37.4 | -0.9 | +7 |
1980 | Kennedy | 2 | 32.2 | 31.2 | -1.0 | -7 |
2016 | Trump | 2 | 28.4 | 24.3 | -4.1 | +2 |
1984 | Mondale | 1 | 53.3 | 48.9 | -4.4 | -18 |
2000 | Bush | 1 | 46.1 | 41.0 | -5.1 | -2 |
1980 | Reagan | 2 | 42.1 | 29.6 | -12.5 | -6 |
Excludes candidates who won less than 1 percent of the vote nationally over the course of the primary. 538 polling averages are calculated retroactively based on our current methodology.
Sources: Polls, CNN, Dave Leip’s Atlas of U.S. Presidential Elections, Iowa Publications Online, “The Iowa Precinct Caucuses: The Making of a Media Event”
First, notice that candidates who beat their polls most in Iowa tended to improve the most in the national polls over the next month. In January 1980, for example, former CIA Director George H.W. Bush won 32 percent of the vote in Iowa, beating his polls by 20 percentage points (Iowa was one of the handful of states that former California Gov. Ronald Reagan did not win that year). The surprise win helped him consolidate support from other candidates and potential candidates — including former President Gerald Ford — and Bush went on to gain 22 points in national polls by the time the New Hampshire primary rolled around in late February.
In the table, you'll also see Obama's and Santorum's Iowa overperformances ranked highly. In both cases, the candidates beat their polls by around 11 points and surged by double digits nationally over the next 30 days.
Three examples do not a pattern make, however. If you look at the top half of the table, you'll find that a good number of candidates who beat expectations gained no ground, or even lost ground, nationally over the next month. But the difference between the bottom and top half of the table is where the story really comes together. See, it's not just that candidates who do well in Iowa gain ground; when candidates do poorly in Iowa, they also tend to lose ground nationally. That's true even if you finish first or second in the primary, as in the cases of former Vice President Walter Mondale in 1984 or then-Sen. Ted Kennedy in 1980.
Measuring this relationship with a regression model, I found that a 1-point difference between a candidate's final polling average in Iowa and their vote share there translates to a roughly 0.9-point change in national polls over the next month. That holds both for increases and decreases: A candidate who crushes expectations in Iowa by 10 points can expect a 9-ish-point gain, on average, in their national polls, while a candidate who falls short of their Iowa polls by 10 points can expect to lose an average of 9 points nationally. I say "on average" because there's still a good amount of uncertainty; some candidates who do well still lose ground, and some who do poorly still make gains.
It also does not seem to matter if a candidate is the winner or the runner-up in Iowa; beating expectations helps them either way. For example, former Sen. Gary Hart was only the runner-up in Iowa in 1984, but his national polling surged more than 30 points in the month after the contest. While the bounce for second-place finishers is smaller on average than for first place, it's not enough to detect with a standard regression analysis (so may just be noise).
DeSantis (and Haley) are still on the back foot
So what does all of this mean for 2024? Well, if past patterns hold, DeSantis's 5-point overperformance would create a 4- to 5-point groundswell in his national support over the next month (though again, that's just an average; his real bounce could be more or less, given the historical uncertainty in the relationship). But DeSantis trails Trump by so much, a 4- to 5-point gain wouldn't be enough for him to give Trump a real scare in any state. In New Hampshire and South Carolina, DeSantis would likely still trail even Haley.
For Haley, who basically matched her polling average in Iowa, the news is a little more mixed. Historically, candidates who perform no better or worse than their polls in Iowa experience no change in their national polls as well. Haley's third-place finish also likely closes the door on a huge bounce before New Hampshire. Worse, if she loses ground nationally and DeSantis gains 5 points, she could come dangerously close to finishing third in her home state of South Carolina, where she currently has a 13-point lead over DeSantis in the polls.
Then, of course, there's Trump. Up over 50 points on Haley and DeSantis in the national polls, the former president is more than a formidable foe — he is basically the presumptive nominee (just without the delegates, yet). He does not need a poll bounce to hold that position. But in fact, he may get one anyway, thanks to an endorsement from Ramaswamy after he dropped out on Monday.
Iowa has held the coveted first spot in every incumbent-less presidential primary cycle this century. Going back even further, polls suggest that what happens there matters. This year, however, it just may not matter enough.
Who will have the momentum out of Iowa? originally appeared on abcnews.go.com