A couple of people have asked me how I predicted Iowa so closely in the face of the final poll from the Des Moines Register poll by Ann Selzer.

“The final Des Moines Register poll was just released, showing Donald Trump leading the Republican field in Iowa with 28 percent, Ted Cruz with 23 percent and Marco Rubio with 15 percent.” source

Selzer famously predicted then-Senator Obama’s caucus victory in 2008, and was the only pollster to do so. This result propelled her from mere pollster to some sort of Iowan Nostradamus, and most pundits and prognosticators ape her projections in the final days. However Selzer’s poll was very wrong in the 2012 Republican Iowa Caucus and I believed it would be again for similar reasons.

The day of the caucus I posted the following:

And the actual results:

The prediction overestimated Cruz and Trump’s support (I thought Carson would underperform), but the order was correct and Rubio’s was right on the money.

In late November I made a slide deck (that I didn’t actually end up presenting anywhere) that argued that Senator Rubio would emerge as the likely nominee after a long, long primary fight. There are a few key points in that deck that explain what happened in IA.

Political junkies and insiders focus too much on polling and not enough on the qualitative dynamics of a campaign. They also have an extremely difficult time viewing the race from a voter’s perspective. Voters don’t follow politics obsessively and make decisions remarkably late in the process. Exit polling in elections consistently bears this out. This is hard for political obsessives to internalize.

• Column One (“Result”): raw result of the question “When did you finally decide for whom to vote in the election?”.
• Column Two (“Undecided %”): the percentage of the electorate up for grabs at the date referenced by the exit poll answer.
• Column Three: The equivalent date of the time in question in the 2016 cycle.

So, for example, a week before the election — January 25 in this cycle — about 40% of the electorate was likely undecided. It actually turned out to be more like 45%, as you’ll see later.

This means elections can be highly volatile, and this is especially true for the Iowa Caucuses. Look for example at the polling in 2012.

This is the Real Clear Politics polling averages in the three months leading up to IA caucus in 2012. Viewed like this, the volatility is really striking. Huge early leads by Cain and Gingrich followed by collapses, and then a final surge by Santorum who was consistently polling sixth until the last few weeks of the campaign. And then came, you know, actual votes:

Santorum was unable to capitalize on this remarkable victory because of a few reasons: confusion about who actually won, his campaign’s decision to wait until too late to give his victory speech — which actually was one of the best of entire cycle — and that he didn’t have a broad enough appeal in the party nationally.

However it was instructive for 2016. Here are the polls over the past few months in IA this time around:

2016 was also volatile

I believed that Cruz and Rubio would over-perform these polls for two primary reasons. One is that religious and ideological conservatives are much more likely to turn out for a caucus.

If you think you are doing the work of the Lord you’re more likely to drudge through the snow to caucus for a couple hours on a weeknight. This played out as over 60% of Iowa Republican caucus-goers were self-described as Evangelical or born-again, They favored Cruz by over 10 points.

I also believed that Rubio would capture late deciders both with his debate performance as well as with his retail politicking. It’s difficult to overstate how talented Senator Rubio is on the ground. Just ask his previous political adversaries. “When Marco Rubio speaks, young women swoon, old women faint, and toilets flush themselves.” That was the Democratic leader of the Florida House Dan Gelber. Or check out this story written after his much-criticized debate performance on Saturday. And, yes, anyone can have a bad night.

These results also demonstrate that in this election, like every other election, there are a ton of late-deciders. 45% of the voters made their decision in the final week.

And that means that almost anything can happen tomorrow in New Hampshire. There are six candidates who could plausibly come in 2nd. Political spectating doesn’t get more exciting than this.