To a certain extent, complexity is a feature, not a bug, of caucuses. And as hard as the Nevada Democratic Party is trying to make Saturday’s caucuses user-friendly, there will still be that pesky multi-step voting process and three different sets of results to wade through. To help you make sense of it all, here’s our guide to how the Nevada caucuses will work — plus how our primary forecast handles all their quirks, and what that might mean for the race moving forward.
Like in Iowa, participants in the Nevada caucuses will physically align themselves with other supporters of their candidate by going to a designated corner of the room. (Caucusgoers can also join an “uncommitted” group as if “uncommitted” were a candidate on the ballot.)
Caucus organizers will then count up the number of people in each group to produce a candidate’s first alignment vote. (In Nevada, this also includes the first choices of people who voted early. This is the first time Nevada offered this as an option, and nearly 75,000 people took advantage of it — quite impressive when you consider that total turnout in the 2016 Nevada Democratic caucuses was about 84,000.)
This is the first of the three sets of election results that Nevada will put out. Think of this first alignment vote like the raw popular vote that primary states report. It’s also what polls of the Nevada caucuses are measuring.
Next up is the famous realignment process. Any candidate that does not meet a given precinct’s “viability threshold” (usually 15 percent of the first alignment vote1) is deemed nonviable, and his or her supporters can thus “realign.” They can choose to join another candidate’s group, or they can simply go home. If their first-choice candidate was just barely nonviable — say, he received 14 percent of the first alignment vote — they can also try to persuade supporters of other nonviable candidates to join their group, in a last-ditch effort to become viable. Voters in viable candidate groupings are locked in, however, and cannot realign.
Realignment is also where early voters’ full ballots will come into play. (Early voters were asked to rank between three and five candidates from their first choice to their last choice.) Because they are not physically present, early voters’ second-, third-, fourth- and fifth-place picks are used to realign them if their first-choice candidate isn’t still in the running. For instance, if an early voter’s first choice is Sen. Amy Klobuchar, her second choice is Sen. Elizabeth Warren and her third choice is philanthropist Tom Steyer, but both Klobuchar and Warren are not viable at her precinct and Steyer is, that voter would be reassigned to Steyer’s camp.
After this process is finished, we are left with the final alignment vote. Notably, this is what our model simulates when projecting a candidate’s expected share of the vote in caucus states. In other words, when our primary forecast says that Sen. Bernie Sanders is expected to win 35 percent of the vote in Nevada, that’s a projection of his final alignment vote.
You can read the details of how our model simulates the realignment process here, but here are the three most important things to keep in mind:
- Other things being equal, the model assumes that larger groups will have an easier time attracting new supporters.
- Conversely, just-barely-nonviable groups will have a harder time doing so in an effort to get over the viability hump. In some simulations of the model, these groups are successful at wooing supporters of other nonviable candidates; other times, they fail to do so.
- Finally, we use a proximity rating to estimate how close the candidates are to one another along a number of ideological and other dimensions. This is a fairly rough method, so our assumptions are fairly conservative — i.e., if Warren voters are trying to decide whether to realign with Sanders or former Vice President Joe Biden, the model doesn’t assume 100 percent of them will go to Sanders, more like about three-fifths of them — but it’s a way to try and gauge where some supporters might go in the realignment process.
After the final alignment, there is still one more set of results to calculate: the number of county delegates2 a candidate has won. Here’s how that works: Each precinct is worth a fixed number of county delegates based on the number of Democrats registered to vote there, and those delegates are assigned to candidates proportionally based on the final alignment vote totals. All precincts’ county delegate tallies are then added together to produce a statewide total; these are the numbers you’re likely to hear most often in news reports. This is because, before this year, county delegate results were the only results released to the public, and county delegates are still the measure used to select Nevada’s 36 pledged delegates to the Democratic National Convention. (I will mercifully spare you the task of explaining the math of that process, but the masochists among you can geek out about it here.)
In predicting who gets a polling boost from “winning” Nevada, our model puts the most weight on the county delegate results — but it also gives a candidate credit for winning the first alignment and final alignment votes. (Specifically, candidate bounces in the model are based on two factors: (1) a candidate’s vote share and (2) a binary variable that indicates whether he or she won the state. We use county delegate percentages to calculate the vote share bounce and will credit up to three “winners” for the binary variable, giving 80 percent to the candidate who got the most county delegates, 10 percent to the candidate who won the final alignment vote and 10 percent to the candidate who won the first alignment vote.)
Of course, this will only matter if different candidates win the three measures. Here’s what our model is forecasting in Nevada not only for the final alignment vote (the number that is displayed publicly), but also for the first alignment vote as of Friday morning:
|Candidate||First Alignment||Final Alignment||Change|
The model currently forecasts that Sanders will receive 29 percent of the first alignment vote, 12 points more than Biden in second place. But it expects Sanders to gain more than 5 points in realignment, pushing him to 35 percent of the final alignment vote. This might be too optimistic for Sanders, though. After all, the day before the Iowa caucuses, the model predicted that Sanders would gain 4.5 points from first to final alignment, but his actual increase was just 1.8 points.3 Then again, if there aren’t any big polling surprises in Nevada, Sanders may be the only viable candidate at many precincts, which wasn’t true in Iowa. That would obviously greatly help his final alignment vote.
As for the other candidates, Biden and former South Bend, Indiana, Mayor Pete Buttigieg barely gain any extra support, and, unsurprisingly, the candidates who are projected to get less than 15 percent statewide in the first alignment are expected to lose support in realignment (they will probably be nonviable at many precincts).
The model doesn’t attempt to forecast how final alignment translates into county delegates, but if Sanders has anywhere close to that 17-point lead on final alignment, he should easily win the most county delegates as well. However, the tendency for Sanders voters to be highly concentrated in certain precincts (like college campuses) can be a disadvantage in the votes-to-delegates conversion, given that precincts are capped at a fixed number of county delegates no matter how high turnout in them soars. So in a closer-than-expected race, it is possible that Sanders wins the popular vote — again — but another candidate is named the “official” Nevada winner by virtue of winning the most county delegates.