Earlier today, my colleague Nathaniel Rakich and I wrote about the final FiveThirtyEight forecast before Super Tuesday. There are basically two big themes there. First, yes, the model has shifted strongly to former Vice President Joe Biden in recent days because of his big win in South Carolina and a major bounce for him in polls of Super Tuesday states. But second, there’s a lot of uncertainty, both in terms of what happens tonight and for the rest of the race. The most likely outcome is still that there’s no pledged delegate majority on June 6, once all states and territories have finished voting.
I’m going to focus more on the uncertainty in this story. Let’s stress test the model and see how its Super Tuesday projections might change if various candidates perform better or worse than our projections expect. (If you want to do your own version of this, you can! Please see our Super Tuesday scenarios interactive here.)
Scenario 1: The model is all-knowing
First, let’s establish a baseline case. Here’s how delegates would be allocated tonight under the DNC’s rules if our forecast is exactly right in every state and Congressional district. If this happens (it will not), I’m going to buy myself a really nice steak dinner tomorrow night. (Also, it will probably mean we’re living in a simulation, so this is not necessarily something to root for.)
In the base case, we have Biden finishing just slightly ahead of Vermont Sen. Bernie Sanders, with 461 delegates (34 percent of the total possible tonight) to Bernie’s 445 (33 percent). But our baseline projections also have Michael Bloomberg and Elizabeth Warren picking up decent numbers of delegates. Heck, they even have Amy Klobuchar staying just above 15 percent of the vote in Minnesota and thereby getting delegates there, even though she dropped out. So basically, the base case for tonight is that one-third of the delegates go to Biden, one-third go to Sanders and one-third go to “other,” which is why we could be headed to a no majority (and possibly contested convention) outcome.
Scenario 2: Biden overperforms
Let’s say, though, that Biden has a really good night — even better than our forecast shows. That’s certainly possible; as aggressive as the model has been in trying to account for his bounce, he keeps moving up in the model every time we feed it new polling data. Even the latest, freshest polls may not sufficiently account for Pete Buttigeg and Klobuchar dropping out and endorsing Biden, which is one way he could beat his forecast, for instance.
To do this, let’s add 6 percentage points to our Biden forecast in every state and district. We’ll take those 6 points evenly from the other major candidates — so Biden gets 2 points each from Sanders, Warren and Bloomberg.
In this case, Biden would finish 169 delegates ahead of Sanders. That’s hardly an insurmountable lead in a race with 3,979 pledged delegates total. But, the narrative of the evening would be quite poor for Sanders, perhaps leading to further Biden gains in subsequent states. That’s because, in this scenario, Sanders’s wins would be limited to Maine and Vermont, with Biden pulling off upsets in Massachusetts, Utah, Colorado and — most importantly, and only by the slimmest of margins — California.
Again, this is a dream scenario for Biden and very much not the base case. But the point is that Biden is surging so much that we can’t rule outcomes like this out. Even in states like California, Sanders’s lead has slipped into the high single digits, to the best we can determine based on the few polls available.
As a slight silver lining for Sanders, note that he actually wins exactly as many delegates here (445) as in the base case. That’s because Biden’s gains would take away votes from Bloomberg and Warren, knocking them below the 15 percent threshold in many districts, leaving Biden and Sanders more delegates to divvy up for themselves. Still, I’m sure Sanders would prefer to avoid this outcome.
Scenario 3: Sanders overperforms
But what if things go really well for Sanders instead? Here’s what the delegates look like if Sanders beats our forecast by 6 points across the board, taken equally from Biden, Bloomberg and Warren.
This scenario has Sanders beating Biden by 118 delegates tonight. A few days ago, this might have seemed like an above-average outcome for Biden — one that left Sanders favored but Biden still in the running.
Now, it might seem like a big moment for Sanders instead. Biden’s wins would be limited to North Carolina, Virginia, Alabama, Oklahoma and Arkasnas — not nothing, but he’d yet to have proven he could win anywhere outside of the South. Meanwhile, Sanders would penetrate into Biden’s territory by winning two southern states, Texas and Tennessee.
Scenario 4: Voters decide it’s Biden vs. Sanders
Next up is a scenario that intuitively seems plausible to me. Let’s imagine voters take the hint that it’s a two-candidate race and gravitate further toward both Biden and Sanders at the expense of the rest of the field. If you add 3 points to both Biden and Sanders’s numbers in every district and subtract 3 points from both Bloomberg and Warren, here’s what you get:
Note that the overall margin between Biden and Sanders has barely changed — but their proportions have improved relative to the field. Biden gets 44 percent of Super Tuesday delegates instead of 34 percent as in Scenario 1, and Sanders gets 43 percent instead of 33 percent. This type of outcome would greatly reduce — though not entirely eliminate — the chance of “no majority” or contested convention. It would also undoubtedly put a lot of pressure on Bloomberg and Warren to exit the race.
Scenario 5: Bloomberg and Warren overperform
Finally, suppose just the opposite happens and Warren and Bloomberg each beat their projections. I don’t necessarily see this happening but who knows! Maybe Warren’s voters are tired of their candidate being discounted and dig in their heels. And maybe Bloomberg’s expensive Super Tuesday ground game has a bigger effect than polls are showing. In accounting for uncertainty in a forecast, it’s important to be prepared for the forecast to be wrong in all directions — not just the directions you intuitively expect.
This would take us far down the path toward a contested convention. Biden and Sanders would essentially tie on the evening with 389 and 388 delegates respectively, but Bloomberg (298 delegates) and Warren (257 delegates) wouldn’t be that far behind. (Note that, because Warren is in danger of failing to meet the 15 percent threshold in more places than Bloomberg is, slightly overperforming her forecasts would be a bigger deal for her than for him.) Nobody would have more than 30 percent of the total delegates.
The tricky part for Warren and Bloomberg is that, although it isn’t that hard for them to accumulate quite a few delegates tonight, winning states is another story. Even in this optimistic scenario, Warren’s only win would come in Massachusetts, her home state. And the best Bloomberg would do would be to tie for the pledged delegates win in a rather odd pairing of states, Arkansas and Utah. I honestly have no idea how such an outcome would be interpreted by voters, the media and the candidates. But it would mean that we were probably in store for a very long and bumpy race.