I have no idea what the result of next week’s Iowa caucus will be. In fact, I have no predictions of any value to make.
The Iowa caucuses of 2020 are one of the most unpredictable presidential primary contests in history, thanks to two unusual dynamics taking place in the same election. First, there are an unusual number of Democrats polling at around the minimum threshold to receive delegates, which is much more complex in a caucus (like Iowa) than a primary (like New Hampshire). Second, the rules of the Iowa caucus have been amended recently to make that delegate threshold much more important. Combining these two elements will make for an unusually volatile race that will be unpredictable no matter how good the polling is.
Many political junkies are aware that in Democratic presidential primaries, a candidate needs a minimum of 15% in order to receive delegates. Win 16% of the vote in New Hampshire and you get delegates – win 14% in New Hampshire and you get nothing. Thus, being between 10% and 20% is a very precarious position. Out of 6 polls taken in Iowa this January so far, Sanders was in that position in 2 polls, Biden in 2 polls, Klobuchar in 2 polls, Buttigieg in 5, and Warren in all 6.
Traditionally, Iowa has not directly reported vote totals. The results reported on Election Day were the aggregated results of hundreds upon hundreds of smaller caucuses in Iowa, which sent the results of each caucus to the County Convention, which then selected the delegates for the District Convention, which chose the actual delegates for the Democratic National Convention. As a result, we historically have not gotten a popular vote result. However, excluding candidates who fail to make viability (such as Martin O’Malley in 2016), the delegates are largely proportional. In 2016, aggregating those small caucuses gave Clinton 49.9% of the delegates and Sanders 49.6% of the delegates, which the Democratic Party used to divide the delegates to each of the conventions. (Clinton eventually got 23/44 pledged delegates).
Critically, in Iowa and many other caucus states, the 15% viability threshold does not simply apply statewide. The threshold applies to every single caucus site (and at some smaller sites, the viability threshold is greater than 15%). A hypothetical candidate who gets 15.0% of the vote in the New Hampshire primary is golden. For a candidate in Iowa with theoretical 15 percent support, things are more complicated. Due to basic statistical variation, they will be above 15% in some caucus sites and below 15% in others, depending on how their supporters are geographically distributed. And at every location where their support is below 15%, their support will be classified as ‘non-viable’ – in other words, effectively at zero. Imagine a candidate who has 15.1% support at half of caucus locations, and 14.9% at the rest. In the final tally, the one the public sees, their support will be just 7.5%. Half of their vote will have been lost! Similarly, a candidate with statewide support of 13% may hit 15% in very few places, and have a final result of just one or two percent. The stakes are clear. For the Democratic candidates struggling on the threshold of viability, a dip of just a few percentage points in public support could be the difference between a respectable showing and a campaign-killing humiliation.
Polls may have a general sense of how caucus attendees vote – but how caucus attendees vote won’t necessarily resemble the electoral outcome! Aware that this is a very strange way to run an election, Iowa Democrats have decided, for the first time ever, to report the results of the first round of caucus voting before any non-viable candidates are eliminated. This provides a chance for a moral victory: A candidate with just 2% of Iowa’s delegates could point out that they did far better among actual voters. But both the Associated Press and the Democratic Party are emphasizing the “delegate equivalent” count, rather than the first-round vote tally, so many voters reading post-caucus headlines about candidate momentum may not immediately notice.
But wait, there’s another wrinkle! Unlike in a primary, in Iowa votes can be changed in the middle of the process. A voter can initially support a “non-viable” candidate, and then switch to another that cleared the viability threshold. But historically, this practice could go the other way as well: A voter could switch from a candidate well over the threshold to get another candidate just on the margins over the hump. Such behavior may sound odd: Why would a supporter of a viable candidate lend a vote to a non-viable candidate they didn’t support? The answer lies in the nature of the caucuses: they are done openly in localities. Democratic voters could see their friends and neighbors struggling to get a candidate over the hump, and lend a hand to support them and keep their votes from being wasted. It may sound strange to Internet political warriors, but most people tend to like their friends that have slightly different political preferences, and thus this phenomenon was common (debating, bargaining, and outright begging often played a role too).
In 2020, though, the rule changed. Supporters of candidates who hit the viability threshold will no longer be allowed to switch to non-viable candidates to help them hit that threshold. Instead, only supporters of non-viable candidates can change their vote. This makes the 15% threshold at each caucus site more absolute, and falling below that threshold far more dangerous for a candidate. But it also invites other forms of chaos: In past years, the rules incentivized votes flowing from strong candidates to those on the threshold of viability. In the new system, the opposite is encouraged: Voters stuck with a candidate just short of viability will pick a frontrunner to prevent their vote being wasted. Moreover, we don’t even know to which frontrunner those voters will flow to. Like the best laid plans, conventional political wisdom tends to collapse upon crashing into real voters. For examples, pundits often assume Warren supporters will all go for Warren and Klobuchar supporters will all go Biden, but real voters are more diverse and complex than dots on a one-dimensional ideological spectrum between the “Goldman Sachs Did Nothing Wrong” caucus and the “Hugo Chavez Did Nothing Wrong” caucus.
The final result? Chaos, confusion, and a totally unpredictable race with just days to go. Candidates separated by just a few percentage points in the first round could have wildly different outcomes: Pete Buttigieg could get 17% of the first round, 25% of delegates, and be a viable national candidate; Elizabeth Warren could finish just 4 percent below him, get 4% of delegates, and possibly be effectively eliminated from the presidential race entirely. Or vice versa. There’s even the possibility of Iowa having “two winners”: A candidate who gets more first-round votes, and a candidate who finishes with more delegates. The confusion about what will happen in Iowa will spill over to the rest of the race. Nate Silver’s electoral model predicts a fairly large bounce for the winner of Iowa (he predicts the winner of the Iowa caucus will be the immediate frontrunner), and we have an incredible amount of uncertainty when it comes to who (or whom) will win in Iowa.
The track record of polls in the last few years have been good. In addition, a lot of smart people with good records have gamed out possibilities, likelihoods, and scenarios. But at the end of the day, this is just a very strange, beautiful process, and I have no idea how it will end. But hey, as weird as it is, electing an American President is still easier than electing a Venetian Doge.
Kirk Jing is an attorney who lives in Los Angeles, California. He wrote for The Dartmouth Review in college before receiving a J.D. from Harvard Law School.