Election polling at a crossroads? Why polling may be on the cusp of big changes

Image 1 of 2

In this election cycle, political campaigns live and die by the latest poll numbers.

If you're leading, they're you're greatest ally. Look no further than presumptive Republican nominee Donald Trump, who repeatedly championed his standing in the polls throughout the primary campaign.

“I’m leading in every single poll,” he said in an interview last summer, as well as countless other times at campaign events.

And if you're behind in the polls, then they're constantly wrong. Bernie Sanders, then-trailing in a New York Democratic primary poll to Hillary Clinton, told me in an interview, “Poll after poll says Sanders can't win - on election day - we do win.”

Good poll numbers fuel fundraising and media coverage, turning candidates into frontrunners. While bad poll numbers can do just the opposite.

But like the campaigns that bad poll numbers throw in to turmoil - some say political polling itself is at a crossroads.

As Courtney Kennedy, director of survey research at the Pew Research Center explains, “A scientific poll traditionally has meant that basically everyone in the population has had a chance of being sampled for the survey. And historically this was very possible by sampling and dialing telephone numbers.”

Pollsters used to be able to use automatic dialers to quickly connect with people on their home phone. But the rise of cell phones has made traditional scientific polling much tougher to do.

A 1991 law called the Telephone Consumer Protection Act made it illegal for pollsters to use automatic dialers to call people on their cell phones. 

That added time, coupled with what Rutgers Public Policy professor Cliff Zukin has said is a rapidly declining response rate and the cost of traditional telephone polling has shot up.

“It's gotten maybe 3 or 4 times as expensive as it used to be just two elections ago,” Zukin said.

And while many see all this amounting to a death warrant for traditional telephone polling - Doug Schwartz disagrees. He's the director of the Quinnipiac poll - one of the most well-known and respected traditional polling institutions in the country.

“Ya know, polls up to this point using telephones have  been very accurate and we see no reason to stop doing it,” he said.

“I don't see any problem with the way traditional telephone are doing what we're doing. If we weren't calling people on their cell phones - we'd have a huge problem. We would be underrepresenting young people - we would be underrepresenting Hispanics - it would be - our polls would be so skewed I would not trust them. But they're not. We're calling cell phones. So we are accurately representing the population and we've been very good at predicting elections.”

On a typical weeknight ,all 150 stations at Quinnipiac’s polling center are filled with pollsters making calls and conducting surveys. It’s a lot of manpower going in to what they say is the gold standard of polls.

But even though the tried and true traditional telephone polling done at Quinnipiac remains the most accurate method of gauging public opinion, the costs can’t be ignored.

“The issue is money,” Schwartz said. “Yes it's expensive to conduct a gold standard poll.”

And because of that, many companies are working to define what the future of election polling will look like.

“Almost all polling will be done on the internet four years from now,” Zukin said.

“The problem with doing internet polling right now is twofold.  First, it's right now not everybody is online, and this is especially true of older people who are most likely to be voters.”

“The second is that there's no good way yet to do a probability sample where everybody has a known chance of being included in the sample of web based users. And so there's a lot of non-probability samples being done this election and those have very little basis in science.”

Now companies like SurveyMonkey and YouGov , as well as big name polling institutions like Pew Research Center, are all working on the secret sauce to have online polls meet scientific muster.

In some cases this can mean using the tried and true dial up method - to bring people to the new online polling world

“We’ve talked to people on the phone and then we migrated them online to do monthly surveys.” Courtney Kennedy said. “That's kind of a nice hybrid where you're getting that online measurement in a timely manner, but we know from the way that we recruited those people that that is nationally representative.”

And once online polling finally hits its stride, some experts say there's really no limit to the kind of information it can provide.

“Instead of being an episodic thing like it is now when it comes out, it will just be there as a data stream. And we'll probably get there by 2020. Certainly by 2024,” Zukin said. Looking beyond 2024, he’s not so sure, “we may have other ways to measure public opinion than calling them up and asking them.”

“Sometimes innovations, even if they enable people, can be disruptive to the way things can be done. And it takes time to test things and build a new body of knowledge and that's where we are now. We're in an experimental phase in survey research.”

And until that day -- it's worth remembering that just because someone is ahead in the polls at any one point - doesn't guarantee a win in November - just ask presidents Mitt Romney, John Kerry and John McCain.