Polls work, but the media doesn’t always interpret them well, pollsters say
Between up and down and October surprises, election polls kept the electorate guessing until election night. But those who conducted polls in 2012 said if one thing could be different, it would be the way polls were interpreted by the media and individuals.
“Polls are being used as substitutes for thinking for a lot of Americans, and that is a really, really dangerous construct,” said Kellyanne Conway, a Republican pollster and adviser for the Newt Gingrich campaign. “The poll is the entire story. It is the headline – it should be the byline in some of these stories.”
Conway said polls should be one element in a larger piece. She and three other experts who spoke Thursday at the National Press Club warned against taking individual polls too seriously without proper context.
“Never ever use the phrase ‘some polls say,’” said John Sides, an associate professor of political science at George Washington University. Sides said a story on a single poll can not only give an inaccurate picture but can also ignore the larger narrative.
For example, in early October, the Pew Research Center poll put Romney ahead by 4 percentage points after showing Obama ahead for most of the race. The Huffington Post splashed the headline across its webpage – it was later updated. But earlier that day, a Gallup poll showed Obama was leading with registered voters by a margin of 5 percentage points. Although the polls were released on the same day, the Huffington Post article notes they were conducted at different times – one included surveys taken after one of the presidential debates.
“Poll averages, poll average, poll averages,” Sides said “The polls are all over the place. You can’t trust the poll someone tweeted at 9 this morning because it’s different than the poll someone tweeted at 9:15.”
Sides suggested using poll averages such as the ones provided by Real Clear Politics for a more accurate picture of the how the country is leaning.
In an email interview Friday, he also pointed to the Washington Post and Pew Research Center as groups that used daily polls but did not report every time a candidate inched forward or backward in the polls.
“They use election polls to talk about more than the basic horse race question,” Sides said. “They explore attitudes toward issues and other facets of the candidates.”
The polls themselves were not the problem, said Mark Blumenthal, who wrote the Huffington Post story about the Gallup and Pew Research polls. He pointed out that most national polls were within one or two points of predicting an Obama victory in the popular vote by two points.
The Internet group YouGov and the Democratic polling firm Public Policy Polling were fairly accurate in saying that Obama would win the popular vote by a margin of 2 percentage points. Outliers such as Gallup were within about four points of the actual number.
Sides had a few basic suggestions. First, if YouGov managed to get the same correct projections as the “scrupulous” Public Policy Polling, then it is time new organizations put more emphasis on polls conducted on the Internet. Unlike traditional polling groups, which reach participants via phone, Internet polling groups use online volunteers, and in the case of YouGov, weight the responses to correspond with the demographics of the nation.
In addition to paying too much attention to polls, journalists also focused too much on issues such as abortion, Margie Omero, a Democratic pollster, said. She traveled the country in 2012 interviewing “Walmart Moms,” women with children under 18 who shopped at the retailer once in the past month.
“Walmart Moms put gay marriage and abortion as seventh or eighth issue they cared about ,” Omero said. “Things like jobs and the economy and education were closer to the top.”
Reach reporter Emily Wilkins at email@example.com or 202-326-9867. SHFWire stories are free to any news organization that gives the reporter a byline and credits the SHFWire.