Next Tuesday, though, things might be a bit more complicated.
The reason? The Iowa Democratic Party will release not one set of data indicating a winner of the caucuses but four. Four opportunities for candidates to frame the results; four chances, potentially, for candidates to declare themselves the winner.
We walked through them on Wednesday, but it’s worth a quick reiteration of how the caucuses work.
Voters first indicate whom they support. Candidates who hit a specified threshold of support (usually 15 percent of caucus attendees) continue on; supporters of other candidates have to pick either one of those viable candidates or band together to make another candidate viable. After that realignment, county delegates are assigned to each candidate. Those are then converted to state-delegate equivalents, an estimate of how much support the candidate will get at the state convention. That support then translates into actual delegates to the Democratic convention this summer, depending on factors like the congressional district from which the delegates come.
That’s just objectively bananas as a process, though it certainly makes things interesting. What’s different this year is that the state party will release not one figure to indicate the outcome — which in the past has been those state-delegate equivalents — but multiple factors. They’ll release the initial level of support for each candidate, the support after voters realign, the state-delegate equivalents and the estimate of national delegates that each candidate will earn.
We were curious how much of a difference that actually makes. How likely is it, that is, that one candidate will win one or two of those metrics and another candidate will win on one or two others? So we built an interactive that models an example caucus.
It’s at the bottom of this article, but we can cut to the chase. Running more than 10,000 caucuses in our interactive with varying polling spreads between the top four candidates, we see a wide range of results. In a scenario where the top four candidates have only a one-point difference between them, nearly half of the caucuses we ran ended up with more than one candidate winning one of the metrics. About 1 in 5 times, three different candidates won. On a few occasions — not many — there were four different winners.
By default, our simulation has a 10-point spread between the top four candidates, the difference exhibited in RealClearPolitics’s average of polls in Iowa on Wednesday. (As we’ve noted, the contest in the state continues to be remarkably volatile.) In that case, weirdness is a bit less common. About 4 out of 5 times, there’s only one winner in our simulation.
Ours wasn’t a full simulation of the caucuses, of course. In reality, there are more than 1,000 caucus locations spread over Iowa’s 99 counties where some 20 serious candidates might be considered by voters.
In our simulation, we scaled things back a bit. We used 10 candidates (represented by AI-generated photos from Generated Media) competing in 16 caucus sites over five imaginary counties. In some cases, we figured there would be favored candidates in particular precincts, maybe because of heavy campaigning or some local connection. The national delegate estimates are based on three hypothetical congressional districts.
The math on this is remarkably complicated, involving models of each individual vote and realignment. An uncommon occurrence in the real caucuses — a failure to hit the viability threshold after the realignment — is common in our simulation, forcing a reliance on the party’s more arcane rules. But the simulation still doesn’t account for the real messiness of the caucuses, like strategic withholding of support or the sudden viability of previously unviable candidates.
The results are still fascinating to watch. And watch you can: Each round of each caucus is laid out before you, to watch at whatever pace you wish. Want to know how messy Iowa could be? Decide for yourself.
How far apart should the four top-polling candidates be?
Caucuses run: 0