On the day of the Iowa caucuses, we devote our weekly Campaign ‘08 Molecast with David Schultz to political polls from Iowa. Our latest conversation with the Hamline University prof and political analyst doesn’t revolve around handicapping the Iowa horse race–we’ll know those answers soon enough–but rather why polling data seems to be growing more erratic and less reliable. (In the last two major surveys of Iowa voters, for instance, CNN and the Des Moines Register differ by 9 points in their assessment of who’s leading the Democratic field and by what margin.)
Polling is fraught not only for new reasons having to do with people’s telephone habits, says Schultz, but for perennial reasons as well. In polling, “how do you figure out who’s actually showing up?” he asks. “Do you ask, do you plan on going to the caucus? Almost everybody will say yes… Often what you see pollsters do is say, ‘we asked likely caucus-goers.’ [But that could mean] people who have gone to the last two caucuses, or people who just said they were going, or…”
On some of the pitfalls of contemporary polling:
“More and more people use cell phones; they don’t use their home phones. Second, a lot of people have answering machines. They don’t pick up their phones…. Those two issues alone make polling less reliable, because we’re not getting a broad, random distribution of individuals. Another problem is trying to decide who one actually samples. The most critical question you want to ask in Des Moines [involves learning] who’s actually going to the caucuses on Thursday night…. How do you figure out who’s actually showing up? Do you ask, do you plan on going to the caucus? Almost everybody will say yes… Often what you see pollsters do is say, ‘we asked likely caucus-goers.’ [But that could mean] people who have gone to the last two caucuses, or people who just said they were going.
“Depending on the methodology you use, you might exclude a lot of individuals. Think about this year in Iowa: Barack Obama is trying to enlist a lot of students and a lot of first-time voters to go to the caucuses. These are individuals who, by most studies, would not be considered likely caucus-goers. Depending on how you survey these individuals, you might get very good results for Obama, or you might get weak results.”
On Iowa’s special caucus rules:
“In order for a candidate to get actual delegates, you have to get at least a 15 percent showing for your candidate in that caucus… If you don’t have that 15 percent, those individuals have to go [to someone] else…. You want to see who people’s second choice is…. When we look at that, we get some very interesting numbers, subject to all the problems of polling: Edwards is the second choice for about 38-40 percent of caucus-goers, Obama comes in the mid-30s, and Clinton comes in very, very far at the tail end. So if you reach a situation where second-choice preferences have to be considered, this could have a big impact on what happens in Iowa.”