The best tea leaf reader in Iowa was the Des Moines Register's Ann Seltzer, which showed Rick Santorum just a percentage point behind Mitt Romney in the last days before the caucus. Santorum finished just eight votes behind Romney, and though no pollster's results showed a tie between the two, Seltzer was the only one who put Santorum in second place. The Register wrote Saturday: "In four days of polling, Romney leads at 24 percent, Paul has 22 percent and Rick Santorum, 15 percent. But if the final two days of polling stand alone, the order reshuffles: Santorum elbows out Paul for second.
Deciding which poll was second best depends on what you value. Picking the winner? Or getting closest to measuring the level of support of the whole field? Many polls picked Romney as the winner -- Insider Advantage, Rasmussen, NBC News/ Marist, CNN/ Time. So did the poll analysis of The New York Times' Nate Silver. But they all show Santorum significantly behind Romney and Paul. Public Policy Polling, on the other hand, wrongly picked Ron Paul as the winner, but showed Santorum the closest to Romney and Ron Paul, with 18 percent of the vote to their 19 percent and 20 percent, respectively.
Though there was some discussion of whether robo-polls -- those conducted through automated phone calls, instead of a real live person on the line -- produced skewed results, human pollsters didn't do any better, Real Clear Politics shows. The NBC poll conducted December 27-28 showed Romney with 23 percent, Paul with 21 percent, and Santorum with 15 percent, while the Rasmussen poll conducted December 28 showed Romney with 23 percent, Paul with 22 percent, and Santorum with 16 percent. Both robopollsters Insider Advantage and PPP showed Santorum gaining in the final days before the vote.