Campaign pollsters pointed to CNN’s likely GOP caucusgoer sub-sample — 544 voters — as too large, given the overall sample of 2,014 adults. Classifying 27 percent of all adults as likely Republican caucusgoers is questionable, they say, since only about one-in-20 eligible voters take part in the caucuses. Their solution: Pollsters should either only call voters with past evidence of participating in primaries or caucuses — or they should tighten their questions to screen out more voters who tell pollsters they will vote but probably won’t.

But public pollsters disagree, arguing it’s too early to know who is going to show up, and who won’t. And they point to the 2008 Democratic contest as a cautionary tale about making assumptions on voter turnout.

J. Ann Selzer, whose surveys for the Des Moines Register and Bloomberg News are considered the gold standard in Iowa, compared the Trump phenomenon to Obama’s surge immediately before the 2008 caucuses. The Obama campaign invested heavily in persuading and mobilizing independents who had never caucused before. Ultimately, three-in-five attendees, Selzer said, were first-time caucusgoers — something that caught then-Sen. Hillary Clinton’s campaign by surprise but Selzer had predicted in her final survey, which showed Obama 7 points ahead of Clinton.