The much-respected Des Moines Register poll was published on Saturday night and shows Mitt Romney in the lead in the Iowa caucuses, with 24 percent of the vote to 22 percent for Representative Ron Paul of Texas and 15 percent for Rick Santorum.
However, as The Des Moines Register’s Jennifer Jacobs noted, there is a “twist” in the survey. Over the final two days of polling on Thursday and Friday, Mr. Santorum got 21 percent of the vote, much better than he polled on Tuesday or Wednesday. Mr. Paul, meanwhile, dropped to 18 percent in the final two days of the survey, while Mr. Romney’s standing remained intact at 24 percent.
Because voter preferences in a primary or caucus can be fickle and can change on a near-daily basis, there is a premium in accounting for the most recent information. Thus, we have entered The Des Moines Register’s poll, which sampled 602 likely voters, as two separate surveys for purposes of our forecasts: treating it as equivalent to a 300-person survey conducted on Tuesday and Wednesday and another 302-person poll conducted on Thursday and Friday. The more recent Thursday and Friday interviews thus receive more weight in our forecasts.
Still, our numbers remain very close to what they were yesterday, as our model had already detected a lead for Mr. Romney and favorable momentum for Mr. Santorum. The model estimates that Mr. Romney has a 63 percent chance of winning the caucuses on Tuesday, Mr. Paul a 21 percent chance, and Mr. Santorum a 12 percent chance.
There is a 5 percent chance of a dark-horse winner — Newt Gingrich, Gov. Rick Perry of Texas or Representative Michele Bachmann of Minnesota. None appears to have favorable momentum in the surveys, but upsets are possible in primaries and caucuses, especially in the first few voting states when voter preferences can be volatile.
Some information in The Des Moines Register’s poll, like the standing for lower-tier candidates such as Jon M. Huntsman Jr., was not included with the initial release of data on Saturday night and will instead be published in the print edition of the newspaper on Sunday. Our forecasts are based upon the information that the newspaper had made publicly available as of Saturday night and they will be updated as these additional details are disclosed.
The Des Moines Register Iowa Poll, run by J. Ann Selzer, is one of the very best surveys in the country. It abides by industry best practices and takes a sample that includes cellphone-only households.
Ms. Selzer is confident enough to let her data speak for itself rather than putting her finger on the scale. On this date in 2007, The Des Moines Register poll showed Barack Obama with a solid lead over Hillary Rodham Clinton and John Edwards, contradicting most other surveys that showed a roughly tied race. But Mr. Obama prevailed in the Democratic caucuses by about the margin the poll projected. No pollster should be judged on the basis of any one election, but Ms. Selzer’s polls have a very good track record over the longer-term as well.
Still, it should be remembered that polls aren’t magic. In fact, there are fundamental constraints on how accurate any poll might be. One of these constraints is imposed by sampling error: somewhere between 100,000 and 150,000 Iowans are likely to attend the Republican caucuses on Tuesday, but a typical poll might survey 400 or 600 or 800 of them.
Another source of error is that polls are conducted some amount of time in advance of the actual voting. This is not always a huge deal in general elections, when voter preferences are locked in early. But in primaries and caucuses — when a great number of voters decide upon a candidate or change their mind at the last moment — taking voters’ temperature even 24 or 48 hours in advance of an election can sometimes miss last-minute shifts in momentum.
Meanwhile, relatively few Iowans turn out for the caucuses and the pollsters can at best make some educated guesses about who will and who won’t.
Even the very best polls, like The Des Moines Register’s, cannot avoid these types of mistakes, so their results should always be interpreted with caution rather than an expectation of perfection. Nevertheless, Mr. Romney and Mr. Santorum should be having an especially happy New Year’s Eve.