Politico has an excellent article explaining just why that is, especially on the GOP side. Three main reasons.
First, there were a very high number of late deciders, more than a third.
The polls missed late deciders.
Iowans considered their decisions up until the last minute. According to the entrance poll, more than a third, 35 percent, of GOP caucus-goers said they decided on their vote in the final days of the race.
But, for the most part, the polls didn’t do the same. That’s not necessarily a bug: News organizations use polls to explain the narrative of the race and for other business purposes. That’s why the Des Moines Register poll comes out on a Saturday night – so it’s ready for Sunday’s newspaper.
Of the polls conducted by live telephone interviewers, only one survey – from Quinnipiac University – was in the field all the way until Caucus Day. But that poll, which showed Trump with a 7-point lead, was conducted over the course of an entire week.
The second-newest poll was the Des Moines Register survey, conducted by Iowa polling legend J. Ann Selzer. But that poll was conducted last Tuesday through Friday, meaning it also may have missed any late movement.
The entrance poll suggests those late deciders went to Cruz and Marco Rubio. Rubio won 30 percent of these voters, and Cruz captured another quarter. Just 14 percent of those who decided in the final days backed Trump.
The last few days in Iowa may have been consequential: Trump skipped the Fox News Channel debate in Des Moines last Thursday. While his threat to boycott the gathering began earlier in the week, that means just one of the four nights of interviews in the Des Moines Register poll were conducted after the debate, for example.
If polls were only meant to predict the horse-race, pollsters would keep interviewing voters until the very last minute. They aren’t, and that’s why the polls may have missed some late shifts.
Second, the polls greatly underestimated the extent of the evangelical vote in Iowa. It ended up exceeding both 2008 and 2012, when it was expected to be far less.
Sixty-four percent of voters identified as evangelical Christians in the entrance poll, higher than in either 2012 (57 percent) or 2008 (60 percent).
These voters went strongly for Cruz: He won 34 percent of the evangelical vote, compared to 22 percent for Trump and 21 percent for Rubio.
The pre-election polls greatly underestimated the evangelical vote. Only 47 percent of voters were self-identified evangelicals in the final Des Moines Register poll. The final Monmouth University poll had evangelicals at 55 percent of the electorate. The Quinnipiac survey out Monday morning was comprised of 39 percent white evangelicals (a minor distinction, given the racial homogeny of the Iowa GOP caucus universe).
Selzer, the Des Moines Register pollster, may have seen this problem coming. The Des Moines Register story on her final poll indicated that the percentage of evangelicals in the survey was lower than historical benchmarks. From the story: “When Selzer rejiggered the new Iowa Poll results to reflect a hypothetical 60 percent evangelical turnout, the race tightens: Trump gets 26 percent of their support, and Cruz gets 25 percent.”
But it was even higher than that.
Third, Cruz beat everyone at turnout. It was assumed that more turnout would help Trump. But it wasn’t his voters that turned out, it was Ted Cruz’s.
The Cruz turnout machine succeeded.
Organization and ground game are difficult to measure before an election. Campaigns tout the number of field offices they’ve opened, or the voter contacts they’ve made.
Using a turnout model that relied on past vote data and stated vote intention, Monmouth University pollster Patrick Murray estimated that turnout larger than the 170,000 voters he predicted would tilt the scales toward Trump.
But record-high turnout, a jump in the percentage of evangelicals and a number of news stories touting the Cruz ground game all indicate that Cruz defeated Trump in Iowa not because Trump’s voters stayed home – but because Cruz identified and mobilized even more of his supporters.
Evangelical voters powered Rick Santorum and Mike Huckabee to victories in the past two Republican caucuses. But they were even more influential Monday night, despite the huge overall surge in turnout.
Fact-based political analysis: awesome. Keep these basic facts in mind as the spin machines go to work over the coming days and weeks. Sometimes a little data can go a long way in understanding just what happened in a particular contest.