Your Mobile Phone Is Killing the Polling Industry
Donald Trump likes to pepper speeches with references to whatever poll puts him ahead of Hillary Clinton, be it nationally or in a battleground state. But any political junkie mainlining cable news in the most divisive U.S. presidential campaign in modern history has got to be wondering why, increasingly, one can find a poll result to fit any outcome.
The lack of consistency isn’t just in this contest. It marred polling in midterm congressional races two years ago and elections and referendums (think Brexit) across the globe. The dissonance can be traced, in part, to the source of the data: the telephone. In the past, pollsters called your landline, so they knew exactly where you lived and perhaps a lot more. They could also dial automatically, setting the machines loose to roll through directories at a fast clip. And more people picked up the phone to say hello, since there was no caller ID. Once on the phone, Americans were more likely to answer questions. Unsurprisingly, back in 1997, the response rate to polling calls was 36 percent.
Fifteen years later, that rate had dropped to 9 percent. Laws protecting your mobile phone from direct marketers largely prevent automatic dialing—calls have to be made manually. And unlike landlines (which many people don’t even have anymore), caller identification is de rigueur on mobile phones, so when pollsters appear on the screen, they’re often swiped into oblivion. Multiply these factors by hundreds or thousands, and you start running into problems.
The solution to this modern conundrum may lie with aggregation, where averages of lots of polls are drawn up to eliminate outliers, but there’s no guarantee that will do the trick. There is, however, at least one silver lining: As more Americans realize polling isn’t as reliable as it used to be, more of them may actually vote.