Over the last year, the accuracy of traditional phone polls, which account for nearly all of the major pollsters in America, has been proven to be extremely unreliable. In an age of do-not-call, unlimited digital voicemail, robocallers, and smartphones, the numbers are proving to be so far off that the standard margin of error can be thrown out the window.
One of the recent examples of this is Greece, a country with similar technological makeup as the United States in regards to landline usage. In July, polls showed that the vote to accept or reject a bailout from the European Commission and IMF was a dead heat going into the final day. The result was a landslide by over 22%, over five times greater than the projected margin of error. Alarm bells started ringing among the pollsters wondering how they could have been so disastrously wrong.
The reason, according to an American pollster who asked not to be named, is that the people willing to answer polls are often less informed and less likely to actually vote. This is a scary prospect for pollsters who seem to be the only ones claiming that their polls are still accurate. The only other defenders seem to be those who benefit from the polls such as current GOP frontrunner Donald Trump.
His campaign points to his popularity as the reason that he’s going to win the nomination, but many are skeptical. James Morris, a U.K.-based partner at Greenberg Quinlan Rosner Research, was defending the poll quality in the United States when he said, “And then there’s Trump. No one knows whether that’s real.”
Doubts over polling intensify after a series of recent high-profile misfires around the world http://t.co/l6c3oZ2X5U pic.twitter.com/Fg1yhsw8rh
— Bloomberg TV (@BloombergTV) October 4, 2015
There have been suggestions that Trump’s popularity in polls is reflective of the demographic that has a landline in their home, answers it, and is willing to take a poll. That’s the magic formula. Fewer people who communicate primarily through smartphones are even willing to take a poll. This is why the participation numbers are falling so dramatically. In their prime, 51%-68% of attempts yielded poll participants. Today, it’s under 7%.
If the polls are dominated by responses from people who are less busy, less informed, and less likely to vote in an election, can they really be trusted?