First, to Iowa. Nunberg says Trump has a natural disadvantage among caucus-goers in the state: a majority of caucus-goers hold college degrees, a demographic which tends not to support Donald Trump. Combined with a net favorability of +17 percent, compared to Sen. Ted Cruz’s net favorability of +55 percent, Trump has very little room to grow.

And while the Trump team might argue that they can win based on bringing newcomers into the Iowa caucus process, there has been no dramatic increase in the number of registered Iowa voters. With just over a month until caucus day, the number of registered Republicans in Iowa (PDF) is roughly that which it was on caucus day in 2012 (PDF), Nunberg said—meaning that while there has been an increase, it hasn’t been nearly large enough to back the Trump campaign’s argument.

“Mr. Trump’s real ceiling is going to be 20 percent on caucus day,” Nunberg predicted, based on what he called a “pessimistic” view of the facts. He pointed to three live surveys of likely Republican voters, conducted by Monmouth, Loras, and the Des Moines Register in December, that show Trump losing to Cruz.

“He’s going to lose Iowa. Cruz will win Iowa,” Nunberg said. “Some of that Trump support will move over to Cruz in New Hampshire, once he loses. I then believe that, because Rubio will… consolidate a lot of that establishment support… he’ll surge at the end.”