Survey Says

How Cruz and Sanders Defied the Iowa Polls, Including Our Own

Bloomberg Politics pollster J. Ann Selzer reflects on how the two candidates beat expectations in Iowa.


Republican presidential candidate Ted Cruz speaks on Feb. 1, 2016, in Jefferson, Iowa.

Photographer: Steve Pope/Getty Images

Last night was a thrill! Yet, as you may have noticed, the outcome of the Republican caucuses was not identical to the final Bloomberg Politics/Des Moines Register Iowa Poll published Saturday. We had the top four candidates identified, all right, but we had Donald Trump with the win over Ted Cruz. 

After the release of the poll—and really, even at the release press conference Saturday evening—we talked about the “what ifs.” What could happen between then and Monday night that would produce a different ending to the story? The caucuses are, after all, designed for surprising twists. 

There is some easy Tuesday morning quarterbacking here. We knew that Cruz held a strong lead with evangelicals. We also knew that evangelicals traditionally show up in higher proportions than the rest of the electorate. Our poll predicted they would be a little less than half of likely caucus-goers. There was no difference in that percentage whether we looked at definite attenders or probable attenders—our way of assessing the impact of turnout. Historically, however, evangelicals have been closer to 60 percent of the Republican vote on caucus night—57 percent in 2012 and 60 percent in 2008. So, we tested a scenario using 60 percent.  That closed the gap between Trump and Cruz to a one-point race, but Trump still led. 

Bloomberg Politics/Des Moines Register Poll Release Event

On caucus night, the unprecedented turnout (about half again as many as showed up in 2012) included a show-rate for evangelicals of 64% in the CNN entrance poll, which was beyond anything in recent memory and what anyone might have speculated, I would guess. Obviously, had we tested that scenario, we would have seen a Cruz victory. We knew Cruz had invested in an aggressive get-out-the-vote effort, with door-knockers and phone-bankers making tens of thousands of late-race contacts. This is probably the biggest reason Cruz prevailed.  

We finished polling Friday night. About a third of Republican caucus-goers, according to the CNN entrance poll, said they made up their minds within a few days of the caucus.  That includes 16 percent who said they did not make up their minds until caucus day—and Marco Rubio got the plurality of that vote (28 percent)—almost doubling the support Trump got with that group (15 percent).

So, Cruz worked Iowa the way Iowa likes to be worked—with a lot of personal touches, both from campaign staff and volunteers. Rubio benefited from late-breaking deciders who felt that he was the best standard bearer, doing especially well with college-educated Republicans and—more importantly—people who care most about nominating a candidate who can win in November’s general election. He swept to a strong third-place in large part because of this group—getting about twice the support (44 percent) as either Trump (24 percent) or Cruz (22 percent) on people for whom this is most important. 

A couple more comments about Trump. He won Western Iowa with 29 percent of the vote over Cruz’s 26 percent. Our poll showed Trump winning the 4th Congressional District by a point. That surprised us, but turned out to be indicative of Trump’s strength in the part of the state most associated with evangelical conservatives. We also showed Trump as the candidate with the biggest share of locked-in supporters. The underbelly of that finding—again from our poll—was that Trump was decidedly unpopular with likely Republican caucus-goers who did not already support Trump. That was not true for Cruz, Rubio, or Carson. People could make a different first-place choice, but still like the other two. But if Trump was not your first choice, you probably did not like him. The obvious opportunity for surprises on caucus night comes from the person-to-person politicking in the room. As people are checking in and waiting for the formal agenda to start, they talk to each other. Trump’s unpopularity made it very hard to convince anyone just leaning toward their candidate or completely undecided to move his way. It proved much easier, on the other hand, for Rubio and Cruz to pick up votes in the room. 

Democratic Race

While our poll showed Hillary Clinton leading, it was by a smaller margin—not the razor thin victory delivered on caucus night. What happened there? 

The biggest difference between the Republican and the Democratic contests was the size of the field. With all due respect to Martin O'Malley, the Democrats had a simpler question to answer—Clinton or Bernie Sanders. The polling showed the Democrats pretty locked down ahead of the caucuses. At the precinct level, that simpler question was handed to O’Malley supporters and the uncommitted. They broke more for Sanders, it would seem—the viable non-Clinton choice. This has always been a contest about which candidate resonates with the mood of the country. Those who had already found Clinton lacking in that appear to have decided Sanders had the better voice for that.

There is also the question of momentum. In our polls, Sanders never lost ground from one poll to the next. He won by substantial margins with younger people and with first-time caucus-goers (with obvious overlap between those groups). The CNN entrance poll shows 84 percent of the under-30 crowd stood up for Sanders. This same age group supported Barack Obama in 2008, but with 57 percent of their vote. 

Sanders won 59 percent of the first-time caucus-goer vote, not far off the margin we showed in our poll. The surprise was how large that group was last night. Our poll suggested it would be around one in three—it was 44 percent. 

Clinton still walked away with 3.77 delegate equivalents more than Sanders. O’Malley took 7.61 delegate equivalents, not even good for 1 percent.

On a personal note, most pollsters are all too aware how dynamic elections can be—caucuses all the more so. We agree to help explain what is happening, and maybe why, as campaigns play out. I believe the Bloomberg Politics/Des Moines Register Iowa Poll contributed a great deal of insight into the two contests. That our final poll did not precisely match the outcome on caucus night bothers me only a little. The reality is I could see how things could shift, and I talked about it almost incessantly in the days after the poll was released. I tell my clients not to get married to early poll numbers while we are still in the field. The warning applies to the final numbers as well. Data can take you only so far. Then, the people speak. What a thrilling journey it was!

Before it's here, it's on the Bloomberg Terminal. LEARN MORE