Princeton Election Consortium

Innovations in democracy since 2004

Sep 27: Biden 353 EV (D+5.2% from toss-up), Senate 52 D, 48 R (D+3.7%), House control D+3.6%
Moneyball states: President IA NV AZ, Senate MT KS AK, Legislatures KS TX NC

How predictive is the Iowa caucus? 2020 edition

February 3rd, 2020, 11:10am by Sam Wang

(Update, Tuesday February 4th 7:45pm: With Iowa returns finally becoming available, it looks like the top four are the same as what I wrote about below. So this post can be read exactly the same – no modification necessary!)

(Pardon the mess – we’re under construction. 2020 content coming soon!)

National surveys and the Iowa caucus aren’t that predictive of a close Presidential nomination race. But the Iowa caucus does drive coverage in the days before the New Hampshire primary, where either the #1 or #2 finisher has gotten the nomination every time over the last 20 years.

Based on analysis of past data – national, Iowa, and New Hampshire – it appears that the nominee will be Joe Biden, Bernie Sanders, or Elizabeth Warren (and maybe Pete Buttigieg.) That’s the case almost no matter what happens in Iowa tonight, though of course Warren has the most to gain or lose. 

And unless past patterns from 2000-2016 get upended, other candidates seem to be out of the money. They may, however, have substantial influence which of those three people the nominee is.

Four years ago, I pointed out the degree to which national, Iowa, and New Hampshire advance polls could predict a major party’s Presidential nominee. The verdict then was Hillary Clinton (D) and Donald Trump (R), and it was clear as of early January.

Back then, I reasoned that election results can be close, by a few points. Combined with polling errors, where parametric approaches (i.e. calculating means, medians, standard deviations, regressions, and so on) may be of limited use, I suggested looking at the data to ask a simpler question: what does current polling rank predict about the nominee?

This year I’m a little late out of the gate since the Iowa caucus is tonight (February 3) and the New Hampshire primary is in eight days, February 11. So I’ll update the table from 2016 to reflect the actual winner of the caucus. The New Hampshire data will be left as pre-primary polls.

Twenty years of primary campaigns (2000 to 2016) show a pattern: in both parties, the eventual nominee always finished in the top two in New Hampshire, and was usually in the top two nationally and in Iowa. Six out of eight nominees, or 0.75 of them, fit the pattern. Here is the data:

Nominee National #1-#2 lead Iowa outcome N.H. polls
2000 (R) G.W. Bush #1 45% #1 (#1)
2000 (D) Gore #1 20% #1 (#1)
2008 (D) Obama #2 19% #1 #1
2016 (D) H. Clinton #1 15% #1 #2
2016 (R) Trump #1 14% #2 #1
2012 (R) Romney #1 8% #2 #1
2004 (D) Kerry #4 7% #1 (#1)
2008 (R) McCain #1 1% #4 #1


For national polls, I show late-December/early-January polls. The “#1-#2 lead” column shows the median difference between the #1 and #2 national candidates. Where New Hampshire polling data was missing, the nominee’s final election outcome is given in parentheses.

In nearly all cases, the eventual nominee has gotten enough attention and support to finish in the top two in all three categories. In this data set, the eventual nominee was at #4 twice, #2 four times, and at #1 sixteen times.

However, things break down when the race is close. The table is arranged in descending order of national margin. Although the Democratic and Republican races in 2000 were nominally open, each had a clear national leader: Al Gore and George W. Bush.

Now focus on Iowa. For races where the national #1-vs-#2 margin was less than 15 percentage points, the caucus is much less predictive. The nominee has always been in the top four finishers – but he/she hasn’t necessarily won (McCain being the outlier).

Let’s look at the 2020 campaign. Here are current standings for Democratic candidates:

Candidate National Iowa N.H.
Biden #1 #2 #3
Sanders #2 #1 #1
Warren #3 #4 #2
Bloomberg #4 #9
Buttigieg #5 #3 #4
Yang #6 #6 #7
Klobuchar #7 #5 #5

In national surveys, this year the top four are Biden, Sanders, Warren, and Bloomberg.

For Iowa, as of today’s final surveys, the top two are Sanders and Biden (near-tied), followed by Buttigieg and Warren (again near-tied).

The intersection of these national-poll-based and Iowa-based datasets is Biden, Sanders, and Warren. So I reckon it will be one of those three. (Or perhaps Pete Buttigieg – he’s at #5 nationally because of the massive ad campaign that has moved Mike Bloomberg to #4.) Of course, this will all have to be revisited after tonight’s results in Iowa.

Finally, one note. Generally, finishing in the top-two nationally and in Iowa has been enough. But since both Biden and Sanders meet that criterion, the rule has to fail at least once this year. Based purely on the data above, New Hampshire will be make-or-break for both of them.

Tags: 2020 Election · President

6 Comments so far ↓

  • Susanne Holland

    Thanks for the analysis. It’s so refreshing, dealing with facts.

  • 538_Refugee

    Bloomberg obviously has the money for polling and strategists. He has decided to skip the first two races and concentrate on Super Tuesday. I can personally tell you he has already flooded this area in Ohio with ads. He has Trump spending money already to counter. If nothing else that’s a plus. Just occurred to me that if Bloomberg is the Democratic nominee, it frees up lots of funds down ticket.

  • Jakob Boman

    Great analysis. My money is on Sanders based on your analysis.

  • Benjamin Rose

    Wondering how you view the impact that “satellite caucuses” may have going forward in other states? Iowa seems to have very recently turned towards Sanders because of the last few percents of prescient reporting, enough to the point where NYT now has fully suspended their “Iowa Caucus Forecast Election Needle” from public view.

  • Fritz Bittrich

    The main impact of IA has never been the how many delegates won but the momentum it causes (or kills.) I wonder if the impact of App Flap can be measured and if that means the NH results are now more meaningful.

Leave a Comment