Buttigieg And/Or Sanders Are Going To Win Iowa. What Happens Next?

Pete Buttigieg and Bernie Sanders each led at least one of the three metrics that Iowa Democrats finally released on Tuesday, reflecting the various ways that Iowa counts its caucus votes. State delegate equivalents, the measure that has gotten by far the most attention from the media because it’s traditionally the way that Iowa has counted its vote, showed Buttigieg ahead 27 to 25 percent, with 62 percent of precincts reporting. But Sanders narrowly led in two measures of the popular vote, taken before and after voters were given the opportunity to realign to a new candidate if their original choice was deemed not viable.

If the split verdict holds, it will be an appropriately weird outcome for a weird-as-hell Iowa caucus. But it also comes at a rather awkward time for our forecast model, which tries to anticipate how polls could change after Iowa. The weirdness makes it harder to guess which direction polls will change in and by how much. (For the time being, the model is still frozen until we get more complete Iowa results.)

The basic way the model works once a state finishes voting is this: It makes educated guesses about how each candidate’s standing in national polls and in subsequent states will improve or decline based on how well they did, accounting for both their share of the vote and whether or not they won the state. This performance is then measured relative to “expectations,” which the model defines as national polls, adjusted by regional factors. The fact that Buttigieg did much better in Iowa (27 percent of state delegate equivalents so far) than his national polls might imply (he’s at only 7 percent nationally) would suggest the possibility of a fairly large bounce, for instance — although the model slightly discounts Buttigieg’s performance since Iowa is a Midwestern state and Buttigieg is from the Midwest.

Our research also finds that much (though not all) of the bounce after major states vote is binary, meaning that who wins the state matters a lot. It isn’t necessarily a continuum. So when there are multiple potential definitions of the winner, that creates a problem.

Before I discuss this further, some good news: None of this is going to matter in a few days. That’s because, once we get a few post-Iowa polls, the model will just use those to measure the bounce and discard whatever assumptions we made initially.

In the interim, however, our assumptions matter quite a bit. In part, that’s because Buttigieg needs all the bounce that he can get to have a shot at winning the most pledged delegates. Based on pre-Iowa polls of New Hampshire, he trailed Sanders 24 to 13 there (and was in fourth place, also behind Joe Biden and Elizabeth Warren). While 11 points isn’t a crazy amount to overcome, especially in a state where polls can be as volatile as they are in New Hampshire, it isn’t easy either — especially when Sanders also performed well in Iowa and could gain more in New Hampshire polls.

And Buttigieg probably needs to win New Hampshire — or come very close to doing so — because the states that follow aren’t good for him. He’s polling at just 6 percent in Nevada and only 4 percent in South Carolina. In other words, it’s highly unlikely, even if Buttigieg does get a big Iowa bounce, that he can win those states (especially South Carolina, given his poor standing with black voters). So he needs to build up enough momentum that he can afford to take losses there and still remain in a reasonably good position for Super Tuesday.

To get an early read on all of this, we ran through 2,500 simulations of each of the following three scenarios on Tuesday night:

• Scenario 1: Iowa results remain exactly as they were initially reported on Tuesday afternoon. In this scenario, we split the winner bonus 50/50 between Sanders and Buttigieg to reflect the competing claims to victory there.
• Scenario 2: Sanders slightly improves his position after the rest of the vote is counted, enough that he wins all three measures and gets 100 percent of the winner bonus.
• Scenario 3: Buttigieg slightly improves his position after the rest of the vote is counted, winning all three measures and getting 100 percent of the winner bonus.

Here are the chances of each candidate winning New Hampshire under each scenario:

As you can see, Sanders is a favorite to win New Hampshire no matter what. It’s really convenient for Sanders that one of his best states in the primary is coming up next. If he’s the unambiguous winner of Iowa, he’d be an 89 percent favorite in the model, whereas if Buttigieg wins Iowa, Sanders would be a 57 percent favorite. Meanwhile, Buttigieg’s chances of winning New Hampshire would be just 4 percent if Sanders is declared the winner in Iowa but would shoot up to 30 percent if Buttigieg wins Iowa unambiguously.

To be honest, that’s probably too wide a spread. Given how long it’s taken to count Iowa’s vote, many media narratives about the race have already set in. Heck, there might not even be much of an Iowa bounce at all if the caucus results get drowned out of the news cycle. Voters may instead simply have a vague sense that both Buttigieg and Sanders did well, and that sense might not change very much depending on who eventually winds up winning each vote count. This is a pretty annoying edge case.

That said, here’s how each of those three scenarios in Iowa outcome could affect the overall chances a candidate has of winning the majority of pledged delegates nationally:

Sanders’s chances would be 64 percent if he’s deemed the winner of Iowa, but only 36 percent if Buttigieg is instead. Buttigieg’s majority chances, on the other hand, would be 10 percent if he wins Iowa — again, keep in mind how much ground he has to make up in more diverse states — but just 1 percent if Sanders wins outright. Again, that’s probably too wide a gap, given the realities of the situation in Iowa and the lack of clear headlines there. And keep in mind that we’re talking about the chance of winning a majority of pledged delegates — and there’s an increasing chance that nobody wins one. If Buttigieg wins Iowa, the chances of no candidate winning a majority of pledged delegates shoots up to 26 percent (they were 17 percent before Iowa). They’d also be at 23 percent with a split verdict.

To recap, our process for updating the model will be as follows:

• Once we feel we can characterize Iowa with a high degree of confidence, we’ll plug the results into the model and turn it back on. If there are still split winners, then both Sanders and Buttigieg will each get at least some credit for winning the state, although we’re still debating internally exactly what that split should look like.¹
• At that point, you’ll see the model swing in one of the directions you see above, depending on which scenario we wind up with. It will also react to all the other new data — new endorsements, polls and so forth — that have not been fed into the model since we turned it off on Monday afternoon. Note that no matter what, Biden is going to have a big downward swing in his chances.
• Finally, once we get post-Iowa polls, the model may react quite strongly to them. We’re expecting Biden to decline in national polls, for instance — if that doesn’t happen, he’ll bounce back to somewhere around where he was before in the forecast (he had a 43 percent chance of winning a majority of pledged delegates before Iowa voted). Maybe Buttigieg shows much more improvement than the model is expecting in Nevada and South Carolina polls. Maybe Sanders, whom the model is expecting to improve his standing, will decline instead.

Any of this is possible — the range of possibilities is always fairly wide after Iowa, and it’s considerably wider than normal because of all the ambiguities there. So you should expect a lot of turbulence in the forecast over the next few days.

Which candidate is spinning Iowa caucuses insanity best? + 538 Has Internships!