scorecardresearch Skip to main content

The heyday of political polls has long passed

Donald Trump spoke with reporters after the presidential debate at the Reagan Library Wednesday in Simi Valley, Calif.Getty Images

As the Republican presidential hopefuls converged for their debate at the Ronald Reagan Library in Simi Valley, Calif., each was hoping to reap political advantage from the national TV exposure. But for most of the candidates who would appear on the stage, any hope of a surge was, at best, a longshot. After all, virtually every poll since the beginning of the campaign had shown the same candidate in the lead: the brash New Yorker with the multiple ex-wives and the take-no-prisoners style.

Is this a description of Donald Trump heading into last Wednesday night’s CNN debate?

Or is it about Rudy Giuliani, heading into the GOP debate at the same Reagan shrine eight years ago?


The fact that it could accurately describe either man in either year tells us a lot about political opinion polls and how crummy they are — crummy not just because of how early it is in the political cycle, but also because of how far we’ve advanced into the digital age.

In the 2016 presidential marathon to date, the Trump phenomenon has been all the rage. Every new survey seems to confirm The Donald’s perch atop the Republican heap. On Tuesday, a New York Times/CBS poll found 27 percent of GOP voters supporting Trump for the nomination, up from 24 percent in August. One day later, a Washington Post/ABC poll had him at 33 percent of Republican voters.

All of which proves exactly nothing about where the presidential race is going.

Eight years ago, it was Giuliani who was the seemingly unbeatable king of the GOP hill. “America’s Mayor” was the hands-down front-runner in almost every political poll — and there were scores of them — taken in 2006 and 2007. For months on end, he dominated the field. Yet in the end he won nothing — not in Iowa, New Hampshire, or South Carolina, and not even in the Florida primary on which he ultimately staked his candidacy. By the end of January, Giuliani had quit.


What was true of Giuliani’s popularity early in the 2008 cycle was true of Howard Dean’s at the same stage in the 2004 cycle. It was true of Rick Perry’s, Newt Gingrich’s, and Herman Cain’s early in the 2012 cycle. Anyone can be the “front-runner” in an election most voters aren’t really thinking about yet. The 2016 New Hampshire primary is still 20 weeks away; the presidential election won’t take place for more than a year. Today’s political opinion polls supply fodder for media pundits and talking heads. But they have about the same predictive power as fortune cookies.

Yet will next year’s polling be any better?

Early poll results have always been rubbish, but election polling itself is growing increasingly dubious. Last May, in a widely-noted blog post titled “The World May Have A Polling Problem,” political statistician Nate Silver confessed that pollsters were in trouble. He was writing in the immediate aftermath of Britain’s general election, when virtually every forecaster, relying on polling data, had failed to discern that David Cameron’s Conservative Party was headed not for a razor-thin plurality in the House of Commons, but for an outright majority — a stunning result.

“There are lots of reasons to worry about the state of the polling industry,” Silver wrote. The UK fail was only one in a string of recent high-profile elections that pollsters had botched. In November 2014, they didn’t detect the Republican wave that swept nine US Senate seats from the Democrats, the largest Senate gain in a midterm election since 1958. Israel’s election in March was thought to be a too-close-to-call battle between Benjamin Netanyahu’s Likud Party and Isaac Herzog’s Labor alliance. In the event, Likud won 30 seats, far surpassing the 18 it had held going into the election.


Election polling is in near crisis,” acknowledges Cliff Zukin, a past president of the American Association for Public Opinion Research. “Our old paradigm has broken down, and we haven’t figured out how to replace it.” The reason? Cellphones.

For decades, professional opinion polling has relied on the ability to call landline phone numbers generated at random and be fairly confident of reaching an adult at the other end. But the explosive growth in cellphone use over the past two decades has fatally undermined that confidence. Today, according to federal data, more than 45 percent of American homes use only cellphones; another 15 percent, though owning landlines, get almost all their calls on cellphones. Thus any pollster who relies on landline phones to survey public opinion bypasses close to 60 percent of US households right off the bat.

Alas, polling firms can’t simply adjust to changing habits. Federal law prohibits the use of automatic dialers to reach cellphones, so pollsters must pay for cell numbers to be called manually — a more costly proposition. Plus, Americans nowadays are rarely willing to take a pollster’s call. Over the course of his career, Zukin writes, telephone response rates have plunged from 80 percent to 8 percent.


Those aren’t the only hurdles tripping up pollsters. Far more Americans now cast absentee ballots, undermining exit polls that rely on interviewing voters at local precincts. Online polling is alluring, but many elderly voters are still not reachable via the Internet.

In short, polling’s heyday is over. Political surveys are ubiquitous, but fewer and fewer of them will be correct. You know that bromide about how the only poll that matters is the one on Election Day? It’s time to start taking it to heart.

Jeff Jacoby can be reached at Follow him on Twitter @jeff_jacoby.