(Editor’s Note: On the day this blog was posted nine polls on the 2016 presidential election were released. Three were revealed the day before.)
On a weekly basis in the United States, pollsters tied to some university, media group or political agenda release their “scientific” take on the 2016 presidential campaign.
Not long after, the airwaves are filled with chatty pundits who will spend the better part of the day deciphering the poll results.
Later in the week, a different set of pundits will talk about how bad political polling has become in the U.S. and lament the credibility of the surveys and their results.
And this will be repeated the following week.
Political polling is fraught with problems, not the least of these is its addictive nature for the media. Polls represent low-hanging fruit that require little effort from journalists — no original reporting and no informational verification. In the business it’s known as a canned story.
Despite all of these inherent problems, Americans love political polls. So, they’re not likely to disappear as the political season stretches through the year, nor will the polls get significantly better.
“It comes down to everyone loves a horse race,” said Dr. Paul Beck, professor emeritus of Ohio State University’s Department of Political Science. “Americans want to know who’s leading and who’s chasing.”
Beck is a longtime academician who once led OSU’s political science department, served as a dean of the College of Social and Behavioral Sciences, worked for a polling firm and has provided expert political analysis for the media.
From his office in Derby Hall he charts polls daily using the website Real Clear Politics. He also scans newspapers — from the New York Times to the Columbus Dispatch — and critiques their polling. He knows polls as well as anyone in America.
And, he’s mostly worried.
“Polling is having serious issues on a few fronts and that’s causing a real concern in terms of their accuracy,” he said. “The largest of these is getting people to respond (which) has dropped terribly to single digits.”
According to most industry officials, cellphone proliferation and the growing refusal to be questioned has greatly impacted pollsters’ abilities to secure enough people for proper sampling.
“Back in the 1970s, it was expected that you could get results from about 80 percent of your sample size. That number has been going down. People just don’t want to take their time to answer polls and they largely distrust them,” he said.
Today, it’s not unusual to get a positive response rate in the single digits. That dwindling sample size is the chief obstacle to accurate numbers. In polling terms, it’s hard to get less than 10 percent survey results and extrapolate that into “what Americans are thinking.”
During a presidential election year, polls become the backbone for a lot of reporting. That’s not necessarily a good thing. Polls can be used as crutches for propping up political reporting that might be better served focusing on issues and not numbers.
“Polls are media darlings because they require no original reporting, especially if it’s your media house who’s paying for the polls. There is little time and resources invested by the newsroom and no need to verify. It represents a perfect package,” Beck said.
As a former editor put it, “It’s like going to a football game and having the sport information people hand you all of the statistics and information. You just write a story off the stats.” (That actually happens in coverage of professional and collegiate sports.)
Press watchdog imediaEthics has consistently harped about the media’s management of polls. The group even gives out awards for the worst in polling.
And then there is the problem of the growing number of polls.
“There’s no doubt that the number of polls has skyrocketed,” Beck said. He points to more and more universities (usually small ones) getting into the polling business and he thinks that’s a mistake. Currently there are 11 schools submitting polling results to Real Clear Politics. Most are not household names. As someone who ran a polling project from Ohio State’s political science department until recently, Beck said the cost of running polls exceeded the value of the questionable results.
“Some universities want to do it for name recognition. This team has a high caliber football program; this one wants to be recognized for its political polling. It’s about getting recognition,” he said. “They want to get the attention when the attention is on politics .… We finally decided (at Ohio State) that we wanted to get out of the polling business. It was expensive and becoming more difficult and we just felt, given the investment, dwindling response rates and the questionable results, it wasn’t something we wanted to attach our name to anymore.”
No one is missing Ohio State’s polling program. How can they? No less than 1,200 polling organizations conducted polls from late 1990s to 2012. They did nearly 37,000 surveys using 3 billion people, according to Harvard history professor Jill Lepore.
But, for all the problems, Beck stops short of a full condemnation. His suggestion is to not put much stock in a single poll. Instead, he likes to use an aggregate of the many.
“There are some polls that I have confidence in, but I think the best gauge is to look at the numbers from a variety of polls and put more trust in the collective results. And, they are surprisingly accurate,” he said.
To read more on the problems with the media and political polls, see these recent stories: