There are six days until Iowa Republicans will head out in the bitter cold (or warmth or snow or rain) to their caucus sites, to argue with other Iowans about which candidate should be the next president: Mike Huckabee or Rick Santorum.

Over the weekend, a Fox News Channel survey suggested that Donald Trump, who'd trailed Ted Cruz at the beginning of the month, surged 15 points to lead Cruz by 11. Well, Quinnipiac University has been polling in Iowa regularly, too, and it found something different. In its poll conducted between Jan. 5 and Jan. 10 (and released on Jan. 11), Quinnipiac had Trump at 31 percent and Cruz at 29 percent. In its poll from Jan. 18 to Jan. 24, overlapping with three days of Fox's most recent poll, the university's pollsters had ... Trump at 31 percent and Cruz at 29 percent.

There were some shifts within demographic groups, but those have much larger margins of error and so probably don't mean much.

Why did Fox see a spike for Trump that Quinnipiac didn't? The difference probably comes down to the methodologies of the two pollsters, as we've noted before. But let's set that aside for now, and simply grant that Trump has a modest lead. (In the Real Clear Politics polling average, Trump is up six points Tuesday morning.)

A caucus isn't like a normal election, though. (Why not? Read this.) It requires a greater investment of time and offers an unusual experience even for people who have voted in Iowa elections before. Which is why this bit of data from that new Quinnipiac poll seems like the most telling.

Among those who've never been to a caucus before, Trump leads Cruz by 13 points — up from the eight points he led among that group two weeks ago. Among those with caucus experience, Cruz leads by three, while the two were tied earlier this month.

Meaning that a greater percentage of Trump's support comes from people who haven't done this before — and who therefore are probably less likely to actually show up. (Studies have shown that voting is a habitual act, and those with a voting history are more likely to vote in the future.) If the race in Iowa is close, that puts Trump at a distinct disadvantage. Trump's experienced supporters are more excited about showing up to vote for him (66 percent say they're more excited in the past to 57 percent saying that for Cruz) — but that's again only among those who've done it before.

There's overlap with another section of Trump's support: lower-income Iowans. Among those making $50,000 or less per year, Trump gets 37 percent support to Cruz's 26 percent. But lower-income people are less frequent voters for a variety of reasons, including that they move around more regularly (and have to update their voter registration information) and often work more unusual hours. (Among the wealthiest Iowans, for what it's worth, Marco Rubio leads.)

It's hack-y for me to say "it all comes down to turnout," which is the sort of thing tedious pundits and desperate campaigns say as elections approach. But with the murky history of primary polls (which have proved to be generally less reliable than general-election ones), with Iowa polls all over the map (thanks to questions about who will turn out) and with the race in that state close ... it's probably true. Trump needs to get his people to the polls, and his campaign knows it.

This poll shows why.