This morning, at a Politico breakfast for hacks and flacks, Obama campaign manager Jim Messina made sport of the pollsters who refused to predict an Obama win. One example: A Massachusetts-based pollster that stopped in three swing states because they were locks for Mitt Romney.
"The Suffolk Poll announced that they were no longer going to poll Florida and Virginia," Messina told Politico's Mike Allen. "What data did you have to say that?"
The man who said it was David Paleologos, representing the Suffolk Poll on the O'Reilly Factor one month before the election. Virginia, Florida, North Carolina -- "We've already painted those states red," he said.
What does he think today? "I should have based my comments on our polling rather than the incumbency rule," says Paleologos. "At that point in time, we had not polled Nevada, New Hampshire, Iowa, or Colorado. Here we are, it's the first week in October, and we don't poll multiple times a week. But the basis of the decision was the incumbency rule. I wanted to get back into those states! We were following that number all up through Monday, we just didn't poll."
The "incumbency" rule suggests that any office-holder polling significantly below 50 percent is looking at defeat. Surely, most of the undecideds are looking for ways to pick the challenger. But the "rule" didn't work this year. To Paleologos's regret, they had some evidence that it wouldn't, and they didn't lead with that.
"We did an 'unlikely voter poll,' asking people who didn't fit the models who they'd support," he says. "I said at the time, there was a disproportionate number of those voters who disliked Romney. Obama did not need to persuade them, he just needed to find them. That poll may as well have been a compass for the Obama campaign. Going forward, we have to understand what the dynamics are for each campaign."