Princeton Election Consortium

Innovations in democracy since 2004

Outcome: Biden 306 EV (D+1.2% from toss-up), Senate 50 D (D+1.0%)
Nov 3 polls: Biden 342 EV (D+5.3%), Senate 50-55 D (D+3.9%), House control D+4.6%
Moneyball states: President AZ NE-2 NV, Senate MT ME AK, Legislatures KS TX NC

Did Iowa turn Biden and Warren into longshots?

February 8th, 2020, 3:09pm by Sam Wang

Last week I pointed out three past criteria that have described the eventual nominee of either party. With the New Hampshire primary just three days away, current polls make it look like two people meet all the criteria: Bernie Sanders and Pete Buttigieg.

After the fuss over difficult tabulation in the Iowa caucus, there was some question of whether it would affect the dynamics of this year’s Democratic nomination. If not, all that fried-butter-eating, or whatever it is that candidates do, would go to waste.

Now we can see the net effect, which is to make Joe Biden’s odds substantially longer.

As I mentioned, in the past, nominees of either party have hit three targets:

  • Top four in national surveys by the time of the Iowa caucus.
  • Top four finishers in Iowa.
  • Top two finishers in New Hampshire.

The pundit talk about Iowa being predictive leaves out the fact that when the race is close, its track record is not that great. Therefore we must count any top-four finisher as a contender.

(Note that Billionaire Mike Bloomberg is at #4 in national surveys, but he isn’t competing in Iowa or New Hampshire. If we leave him out, that puts Mayor Pete Buttigieg at #4.)

The New Hampshire primary’s not until this coming Tuesday. But at this point, the opinion polls are starting to coalesce.

If we use them to guess the top two finishers, what does it all look like in a Venn diagram?

There are two candidates in the intersection of all these sets: Senator Bernie Sanders and Mayor Pete Buttigieg. If the pattern of 2000-2016 is followed, the Democratic nominee will be one of these two.

Past processes are running more or less smoothly, in the sense that they are producing at least one viable candidate. Bloomberg might have trouble getting further traction, unless there’s some kind of deadlock where everybody turns to him. However, Warren is already available for that: preference polls indicate that she is acceptable to the highest fraction of Democratic voters.

From a pure Venn-diagrammatic point of view, the Iowa results don’t add information. But they did lead Biden and Buttigieg to switch places in the New Hampshire race. In that respect, the net effect of the Iowa caucus was to make it much harder for Joe Biden to get the nomination.

This leads to the question that’s come up in Democratic Party circles, which is whether to give such outsize influence to a state like Iowa that does not reflect the party’s diversity. It’s easy to make an argument that it would matter.

Early election results appear to give a bounce to the top 1-2 finishers, as voters in other states find out which candidates perform well in an actual election (or caucus, in Iowa’s case). If the first primary were in South Carolina and not Iowa, then presumably Joe Biden would have gotten the benefit. I would be writing about a Sanders-vs-Biden faceoff. In any event, the Iowa caucus favored the moderate liked by whites (Buttigieg) over the moderate liked by blacks (Biden) – and it’s affecting New Hampshire, a fairly predictive state.

If some other candidate eventually gets the nomination, it would identify forces not captured by the simple rule I have described. Biden had strong support in South Carolina (primary February 29), and Warren was performing in the top four in national surveys. The normal course of events at this point would be for them to decline in support. This should all become much clearer on Super Tuesday, March 3, the date of 15 more primaries nationwide.

What’s the probability of an exception to the pattern? 2000-2016 gave only 8 examples, a pretty short record. I would guess the probability of an exception (i.e. someone other than Buttigieg or Sanders) to be about 1 in 10.

Tags: 2020 Election · President

3 Comments so far ↓

  • Andy

    I think Biden has been ripe for a polling drop-off no matter how the states line up. As it is, South Carolina might rescue him; it’s hard to imagine Biden voters moving en masse to Buttigieg, for example.

    On the other hand, if South Carolina had been first, maybe we’d still be looking at Harris and/or Booker as the probable beneficiary of a Biden collapse.

  • bks

    Sanders was the victim of his own device as he received a plurality of the vote in Iowa, but was denied the immediate media coverage by the “app flap”.

    Klobuchar should be the recipient of the Biden voters, but Bloomberg is rising from the deep like Moby Dick.

    The knock on Sanders is that he can’t get the minority vote, but in fact he polls extremely well among Latinos (Arizona is in play, and maybe Texas?)

    Put me down in favor of a contested convention which would provide riveting media coverage for three or four days with great drama.

  • AP

    Interesting. 538 has, for the first time, a primary model and they have Biden in the top two, albeit much diminished. Sanders is 3 in 5 to get a plurality at the convention. Buttigieg still much lower. The model is much more complex than the criteria used in this blog, not necessarily better, thus I am not sure where this big difference in prediction comes form. My guess is that Buttigieg is in 4th position nationally by quite some margin. It’d be nice to put all forecasters in a mud pit … I mean a nice podcast studio and have them fight it out … I mean explain to the public why forecasts differ and what the past record suggests. Then have a post election meeting with edible bugs available ;)

Leave a Comment