(Weekly Updated Map) Fivethirtyeight's Primary Forecast

[Horatii]

January 9, 2020

Delegate Forecast:
Biden: 1,512
Sanders: 1,000
Warren: 643
Buttigieg: 538

Obviously this is quite bullish on Biden, but it'll be interesting to have a record of how it changes over time. I'll update the map every week or so.

https://projects.fivethirtyeight.com/2020-primary-forecast/

[ProudModerate2]

Can you provide a quick link to the original source.

[Crumpets]

Can't believe he has WY at 30+% Biden when we already have poll showing Warren up by 80 there.

[Horatii]

Can you provide a quick link to the original source.

Added a link to the original post.

[Senator Spark]

Ridiculous. Hope it will change as we go along.

[538Electoral]

Dang.

[Libertas Vel Mors]

Ridiculous. Hope it will change as we go along.

It's actually perfectly reasonable lol, just expecting big swings based on Iowa. This current map has Biden leading the polls in Iowa, which of course if he wins he will cruise onwards, as this forum has agreed as well. If you think he's going to lose Iowa, of course the map will be different, but it's dumb to criticize the model.

[Mr. Morden]

The model assumes dynamic interaction between results of the primaries in sequence, so that early primary outcomes can influence later primaries.  So, for example, it currently gives Biden at least a plurality probability of victory in 3 of the first 4 states (IA, NV, SC), and Sanders a plurality probability of victory in NH.

Here are the current chances it’s giving of each of the top 4 candidates winning Iowa, for example:

IA:
Biden 33% chance
Sanders 27% chance
Buttigieg 22% chance
Warren 14% chance

Biden’s also of course ahead in NV and SC.

So in the median case, Biden wins 3 of the first 4 states, and heads into Super Tuesday with momentum.  And the model currently thinks that momentum is just enough to make Biden at least a narrow favorite in all of the Super Tuesday except Vermont, including states where he'd otherwise be the underdog because of current polls or demographics.  But many of the non-Southern Super Tuesday states are really close.  Here are the Super Tuesday states in which Biden’s “lead” in probability is less than 5 percentage points:

CA:
Biden 34% chance
Sanders 32% chance

MA:
Biden 31% chance
Warren 28% chance
Sanders 24% chance

MN:
Biden 29% chance
Sanders 25% chance

CO:
Biden 34% chance
Sanders 32% chance

UT:
Biden 34% chance
Sanders 33% chance

Washington is another state that’s very close, though it votes later than Super Tuesday:

WA:
Biden 35% chance
Sanders 32% chance

The model thinking that Biden is the most likely to win Iowa is largely based on the fact that there’s been very little Iowa polling recently, but there’s been pro-Biden movement over the past six weeks or so in national polls (or at least negative movement for Buttigieg and Warren).  So it thinks Biden has gained in Iowa too.  But tomorrow we have an A+ pollster coming out with a new Iowa poll (Selzer).  So if that poll shows Biden trailing in Iowa, than many of the states I mention above could conceivably tip from narrow advantage Biden to narrow advantage Sanders (or Warren in the case of MA).

[Figueira]

Biden is the plurality favorite to win all the later states because all the other candidates are still in the race. If it becomes a 2-person race, the non-Biden candidate should be able to win at least some other states. Remember, 538 doesn't even give Biden 50% odds to win a plurality of pledged delegates.

[Crumpets]

That DMR poll has already had a pretty big impact on the forecast:

[Mr. Morden]

In the 2016 primary, the results in the South were so lopsided in Clinton's favor that a hypothetical Clinton-Sanders national pledged delegate tie map would have had Clinton still winning every or nearly every Southern state and Sanders sweeping virtually every non-Southern state.  So in that hypothetical world, there would have been a delegate tie, but Sanders would have won many more states.

From browsing through this model, it looks like, if we were to take it at face value, then I don't think a Biden-Sanders pledged delegate tie would look quite the same.  It would still be very Southern-heavy for Biden, but it would be ~25 states won for each candidate, in no small part because while Sanders is strong in a lot of small states, one of his strongest in the country is delegate motherlode California.

[Crumpets]

Well, if nobody else is going to update this, I suppose I will.

[Xing]

Instead of stating my feelings about the current model, I'm wondering how actual results will change the model. If Sanders wins IA and Biden gets 3rd, will they still have Biden beating Sanders in 47-48 more states, including NH? Or will that shift their projections in the remaining states?

[Non Swing Voter]

If Biden actually wins Iowa it's somewhat plausible...

[Mr. Morden]

It’s currently got:

Iowa:
Biden 30% chance of winning
Sanders 29% chance of winning
Buttigieg 21% chance of winning
Warren 18% chance of winning

The rest of the forecast is very sensitive to the projected Iowa results.  Following the NYT poll of Iowa being released this morning, it showed a clear boost for Sanders in the rest of the states.

Sanders is currently the favorite in:
NH, NV, CA, MN, CO, UT, ME, VT, WA, Dems Abroad

While Warren is favored in MA, and Biden is favored everywhere else.

Note that Biden is favored in every state that votes later than March 10th.  I guess that’s because the model figures that whoever’s leading will most likely be far enough ahead by then that subsequent contests will be uncompetitive.  And since Biden is still favored overall, they still have him favored in all of the states with later primaries.  But if Sanders gets some more Iowa polls like today’s then there could be a big shift in his direction.

[Mr. Morden]

https://twitter.com/NateSilver538/status/1221502594260049926

[Holmes]

Funny that his model has NY going to Sanders but not MI in a tied race.

[Libertas Vel Mors]

Biden red, Sanders green

[Epaminondas]

That maps looks ripe for a wave of green to wash over it next Tuesday.

[Mr. Morden]

Warren seems to have been tanking in the model recently, now down to 7% chance of a pledged delegate majority, barely above Buttigieg at 6%.  She's not eve in the top 2 in Massachusetts anymore.  Sanders and Biden are both rated as more likely than her to win the state (presumably because the model figures there's a good chance she'll be out of the race by then).

I assume part of this is due to the model accounting for "lanes".  Sanders doing better hurts Warren, because they have voter overlap.  But I'm guessing there's also the fact that, as time goes on with Warren and Buttigieg both lagging behind Biden and Sanders nationally, the model doesn't see much path for either of them to catch up unless they manage to get a big boost out of wins in IA and NH.  But the polling average has Warren in just 4th place in both states, while Buttigieg is in 3rd, so Buttigieg has a better chance of pulling off the upset in one or both states.  If that doesn't change, we could soon see Warren drop to 4th place in the overall model even while she remains ahead of Buttigieg in the national polls.

[GeorgiaModerate]

Biden red, Sanders green

For aesthetic reasons, either Oregon or Idaho needs to flip.

[Crumpets]

Bumping this thread with an updated map: