“We’ve seen the polls wrong last night all across the country,” said CBS-6 political analyst Dr. Bob Holsworth. While many of the upsets were within the statistical margin of error, many others – Virginia, Illinois, Maryland, for example – weren’t.
Geoffrey Skelley with University of Virginia’s Center for Politics said “the fact that it missed in a lot of different places would suggest there’s something more to this . . . Pollsters are faced with, if not a crisis, they’re moving in that direction. It’s getting harder and harder to do what they do.”
And it’s been a notable issue the previous two election cycles.
Bad polls can actually change the result by discouraging voter turnout and shift campaign spending.
Polls are also a crucial way of gauging what citizens are concerned about, and therefore, where politicians focus their energies.
Both Holsworth and Skelley believe one reason polls have slipped in accuracy is that the old way of polling by phone isn’t getting a strong enough sample because more and more people don’t have land lines and don’t answer cell phone calls from strange numbers.
Plus, Skelley said, it’s against the law to place automated poll calls to cell phones. Holsworth also believes it’s getting harder for pollsters to get a solid feel for whether the person they’re polling is actually going to vote.
More and more political activists accuse even non-partisan pollsters of having a bias, which may be fueling distrust of pollsters and, perhaps, encouraging more respondents to be less that truthful. And with smaller, perhaps more questionable samples, the sweet science of extrapolating the results becomes more risky.
Pollsters don’t want to get it wrong. Their accuracy is their credibility.
So, they’re going to have to get creative: Tap into social media, meet people where they’re comfortable – a polling app, for example.
This last cycle there was more online polling, Skelley said. For example, the New York Times partnered with an internet polling firm. “I don’t know if anyone has done a deep dive just yet on how effective that new form was. But it’s probably the direction that pollsters are going to have to go,” he said.
Hollsworth said “the notion of polling being done by just sending out some questionnaires, getting a phone room in which 300 people, or 50 people, are going to be calling 500 people in Virginia, I think we’re going to see that go by the wayside in the next 10 years.”