The Sanders campaign is sending out mailings telling supporters that they’ve still a little behind but they’ve almost caught up, and would you send us just $3.00 so we can win? The Clinton campaign is telling people on its website that “the polls are tight, and this is anyone’s game,” but we can win if you help by sending a prepared message to your friends in Iowa.

The Clinton campaign’s materials don’t mention Sanders. They do mention Obama (she’s his heir) and the Republicans (she’s the only hope to save the country from them).

Sanders does mention Clinton: “When we first launched our campaign, the polls in Iowa showed us down 55 points against Hillary Clinton. The political media laughed at us. They said our ‘radical ideas’ would never attract much support against an ‘inevitable candidate.’ But together we’ve built a campaign on the verge of accomplishing what everyone but us thought was impossible: winning the Democratic nomination for President of the United States.”

The New Yorker explains: “With a week to go until the Democratic caucuses in Iowa, it would take a brave person to bet on the outcome. The opinion polls are all over the place, and many of them have technical issues. If you average out the most recent surveys, Hillary Clinton retains a narrow lead [but] Bernie Sanders is drawing larger crowds to his events, and he appears to have momentum on his side. . . . An old adage among longtime observers of the Iowa caucuses is ‘ignore the polls and look at the ground games.’”

The magazine’s writer isn’t sure what the ground games say. Clinton organized better and much earlier than Sanders, but Sanders has almost caught up in formal organization and seems to be doing better rallying voters on the ground. But journalists can only report what they see and in looking at a candidate’s “ground game” they don’t see enough. So everyone relies on polls, which claim to be scientific, but even the polls don’t help much this time, because they disagree so widely.

The polls range from one giving Clinton a huge lead (59% to 30%) to another giving Sanders an 8% lead (51% to 43%) among those likely to vote. The New Yorker explains this too: much of the difference comes from how the polls decide who is likely to vote. The looser the definition, the more likely the poll is to show Sanders doing well. So:

Clinton still appears to be ahead among Democrats who have participated in previous caucuses. Sanders is pulling in younger Democrats, independents, and folks who haven’t been to a caucus before. Next Monday, the turnout among these groups will probably determine the result.

In the 2008 primary, reports the International Business Times, Clinton led Obama in the polls until the last week. “When she lost to him that year, it was one of the first major signs her campaign was in trouble. This time, her national lead over Sanders has dropped much more quickly than it did over Obama.”

The experts disagree

The difficulty in predicting Iowa isn’t just the polls disagree but that the experts disagree on what the polls mean. The IBT declares that “An analysis of historical polling data shows that even less than one week from the first voting of 2016, virtually anything can happen in Iowa.” The way the causes work — the effort involved in participating and the personal nature of the caucuses —mean that votes could change at the caucuses themselves. The newspaper points to the 2012 Republican caucus, when Rick Santorum, presumably boosted by a strong vote from religious Iowans, went from a very distant sixth place a week before the caucus to win over the favored Mitt Romney.

The statistics-nerd site FiveThirtyEight disagrees, with qualifications. “It’s true that even the final Iowa polls are sometimes way off,” writes Harry Enten. But, he says, within ten days of the caucus they give a good rough idea of how each candidate will do. Reviewing the polling and the final results for every caucus starting in 1980, he notes that every winner but one “was either within about 10 percentage points of the leader or showing at least some momentum in the polls by this point.”

That leaves a lot of leeway, of course. In the 2004 Democratic primaries, for example, “John Kerry was in third place in the polls with 16 percent, while Howard Dean was leading the field at 27 percent. John Edwards averaged just 11 percent a week before the caucuses, but that was up from 5 percent in December. The final result: Kerry 37 percent, Edwards 33 percent, Dean 17 percent.”

And all he’s say, which is all any of the polling experts will say, is that he’ll be shocked if someone other than Clinton or Sanders wins the caucus. This we don’t need pollsters to know.

Real Clear Politics provides a page with graphic picture of its poll average — which as we write shows Clinton ahead 46% to 45.4% — and a table with the results of the major polls since last May.