Use a careful eye with political polls
In the throes of a tight and turbulent election season, insight into how a particular candidate is faring is a hot commodity among campaigns and political wonks alike. Enter political polling — equally ballyhooed and maligned, but always of great interest.
Polls are intended to give us a snapshot in time of how a race is shaking out, but as the 2016 presidential election exposed, it’s buyer beware when it comes to depending on these public surveys as election predictors.
Last week the Greater Montana Foundation along with the UM Bureau of Business and Economic Research, offered Montana media an opportunity to hear from a panel of experts about public opinion research and how to decipher good polling from bad.
“There have been lots of conversations about the accuracy of polling,” said Nicole McCleskey, with Public Opinion Strategies. “I’ve even heard the question, ‘Is polling dead?’”
While it’s not dead, she said, it is changing. Consider this: In 1997 a survey conducted over the phone generated a 36% response rate. Today, phone polls only garner about a 6% response. Thanks to smartphones, anyone can block calls or simply ignore unknown numbers. In turn, telephone surveys have become much more difficult, and more expensive.
Today, technology is driving the industry. But as McCleskey points out, while useful, an internet-driven survey has drawbacks. She notes that overall 10% of Americans are totally off-line, which automatically eliminates a vast subgroup of voters. That rate is even higher in Montana.
It’s also more difficult to get a truly random sample of the population online.
“I come from a view that not all polling is created equal,” McCleskey said.
So what should we look for when considering the accuracy of a political poll?
Along with methodology, it’s important to consider who funded the poll, the sample size (surveys with more than 500 respondents are preferable), the demographic and partisan composition of the survey, and how the questions were presented.
The final numbers matter, too. If a poll result falls within the margin of error, it should be considered too close to call.
Transparency is also key. A good pollster will make all of this information readily available.
Dr. Eric Raile, a political science professor at MSU, suggests looking at multiple polls for a more accurate snapshot of an election.
“Consider each poll as one data point,” he said. “Put some credence into each poll, but look at what’s happening globally.”
And take it all with a grain of salt.
As the Associated Press Stylebook warns, “no matter how good the poll, no matter how wide the margin,” a poll does not say one candidate will win an election. “Polls can be wrong and voters can quickly change their minds before they cast their ballots.”
What’s in the cards for a candidate today may not be the same come November.