As some pundits portray it, national pollsters have not fared so well this year proving the accuracy of their craft and things may just get worse as the November midterm elections roll around.
After a string of surprise victories in this month’s primaries, political observers on both sides of the ideological divide are accusing national polling firms of misreading this year’s voters.
“I think the polling models this year are all wrong,” said conservative commentator George Will on ABC’s “This Week” last Sunday. “The polls are not picking up the change in the turnout and the composition of this year’s electorate.”
Will cited Alaska’s ongoing Republican Senate primary, in which underdog candidate Joe Miller could upset eight-year incumbent Sen. Lisa Murkowski when all the votes are ultimately tallied from last Tuesday’s election. While Miller retains a slight lead with most of the votes already counted, polls before the election showed Miller to be trailing Murkowski thirty points. In Florida, some national pollsters predicted that state Attorney General Bill McCollum would trounce businessman Rick Scott in the state Republican gubernatorial primary, but Scott defeated him by three percentage points.
“George is right,” said Democratic strategist Donna Brazile. “There’s an undercurrent right now in the country of voters who are simply disgusted with the politics — what I call the status quo politics. And it caught…Senator Murkowski by surprise. It will catch many other folks by surprise this fall.”
But pollsters contend that while 2010 certainly has a unique flavor, their reading of the data is not nearly as bad as the pundits are making it out to be.
One of the major challenges of successful primary season polling involves making an educated guess about who is and who is not likely to vote, so some races are bound to have surprises, they said. Each election cycle is different, but 2010 requires a bit more maneuvering than in years past given the types of people who only recently felt motivated to make the trip to the election booth.
“My instinct is that this year you’re seeing folks who are real energized to turn out in a way maybe they haven’t been in previous elections,” said Kristen Soltis, director of policy research at the Winston Group, a conservative strategy and polling firm. “That is what’s making it difficult to model these elections.”
While the public’s attitude toward politics is in constant flux, it’s not as though 2010 marked a slow shift in which the American people gradually lost favor with one party and switched to the other. Much of the country transferred their frustration from Red to Blue in near record time, pollsters said.
Tom Jensen, director of the Democratic firm Public Policy Polling, called it a “cataclysmic drop-off from 2008.”
“It’s possible that the difference between people that showed up in 2008 and 2010 will be more drastic than we’ve ever seen,” he said. “And that’s something that pollsters are going to need to pick up on and account for.”
While that may partially explain why it is growing more difficult for pollsters to define “likely voters,” one other factor to consider is that at no other time has there been so much polling done for a midterm election. When the Republicans took control of Congress in 1994, for instance, public polls for primary House races were rare, and many of the polling firms disseminating surveys today did not even exist then. The proliferation of 24-hour cable news channels and the insatiable thirst for data the Internet supports have created an environment where polling information can be spread across the country at any time, a reality that has enticed more people to get into the polling business.