When Hillary trounced Trump: Polls are unreliable a year before an election. So why are Dems watching?

Hillary Clinton had a commanding lead over Donald Trump. John McCain and Barack Obama were in a virtual dead heat. President George H.W. Bush was crushing Bill Clinton.

So polls said in 2015, 2007, and early 1992 — months and months before the respective presidential elections.

Democrats desperate to beat President Donald Trump this time around have been focused on one word looking toward 2020: electability. And polls should provide a kind of guidepost. But if history is any indication, polling more than a year away from a general election is terribly unreliable.

“They’re virtually useless,” said Sarah Niebler, an assistant professor of public opinion and campaign behavior at Dickinson College in Carlisle, Pa. So far from the presidential election, “the polls you see that match up Trump vs. Biden or Trump vs. Warren or whoever, those have zero explanatory power. They make political scientists crazy when polls are held up to say, this is the person who can win."

Polls are a public opinion snapshot, but they are far from reliably predictive more than a year before ballots are cast. Primary horse-race polls tend to be somewhat more reliable, with some notable exceptions: Jeb Bush and Ben Carson were at different times favorites in the Republican primary campaign of 2016 until they weren’t; Herman Cain led the pack of Republicans vying for the 2012 primary in October 2011 polls.

“You have to understand, most people are not paying attention, and a lot of what we’re picking up is name recognition only,” said Krista Jenkins, professor of politics at Fairleigh Dickinson University in North Jersey and director of the FDU poll.

But winning seems especially important to Democrats this year. Selecting a candidate who can beat Trump, Democratic voters say, is a priority over issues, policies, ideology, or nominating a woman or racial minority, according to recent surveys.

“Democrats want an electable candidate,” said Patrick Murray, director of the Monmouth University Polling Institute. “And then, when you talk to them about what they think that is, it’s basically what the media has been saying about their chances.… That’s a key part of Joe Biden’s support right now. A large number of voters are saying, ‘I keep hearing that he’s electable.’ In many ways, it’s a self-fulfilling prophecy. They’re concerned about electability, but we’re not in a position to judge that yet.”

Despite their relative unpredictability, polls are sought out by voters seeking signals on whom to support. Candidates who poll well have historically been shown to get more media attention, and the reverse is true: The more media attention a candidate gets, the more likely the poll numbers are to rise.

Polling also plays a part in determining who gets media attention via the debate stages. To participate in the July 30 and 31 debates, candidates had to get at least 65,000 donations, or at least 1 percent in three polls recognized by the Democratic National Committee. That threshold goes up to 2 percent for debates in September.

“The media is treating polls today with the same kind of reliability as they would have in late October of the election year,” Murray said. “The problem is, these early horse-race polls don’t tell us what they’re portrayed as. They don’t tell us how strong is Joe Biden relative to Sanders or Kamala Harris against Trump. They don’t tell us that because the voters don’t really know the difference between Kamala Harris and Joe Biden yet. They don’t know what kind of general election race they would run and what kind of attacks will come from Trump.”

Most general election polls show Biden having the widest lead on Trump in head-to-head match ups. An NBC/Wall Street Journal poll released last week showed the former vice president leading Trump by 9 points, Sen. Bernie Sanders beating Trump by 7, and Sen. Elizabeth Warren beating him by 5 points.

Historically, general election polls more than a year out are about 11 points off, according to a FiveThirtyEight analysis of general election polls from 1944 to 2012.

Take for example late 1991, when Bill Clinton looked like toast, trailing President Bush by 21 percentage points. Then, after the Gulf War ended, the nation focused more on the economy. Bush’s favorability waned and Clinton won.

Polls had Jimmy Carter beating Ronald Reagan by 16 points a year away from the 1980 election. Reagan went on to win by 10 points.

Lyndon Johnson led Barry Goldwater by 50 percentage points, and went on to beat him by 23. It was still a blowout, but less than half of the margin those early polls predicted.

More recently, a year out from the 2016 election, a Quinnipiac poll had Ben Carson beating Hillary Clinton by 10 points.

“We think a country 200-plus years old, we’ve had all these elections, we should be really good at predicting this stuff, and it turns out, we don’t have that many and we’re not,” Niebler, of Dickinson, said.

Polls also tend to be conducted nationally, whereas a presidential race is determined state by state. State polls (who would win in Pennsylvania between Biden and Trump, for instance) might offer slightly more insight into a general election victory but should still be read with skepticism this far out, said G. Terry Madonna, professor of public affairs and director of the Center for Politics and Public Affairs at Franklin and Marshall College. (A Quinnipiac poll released Thursday showed Biden as the only Democrat who beats Trump in Ohio.)

A candidate’s perceived electability isn’t just based on polls. Biden has positioned himself to voters as the person with the best chance to take on Trump because of his background and politics. He touts his roots in Scranton as a path to take back Pennsylvania for the Democrats. And he appeals to Rust Belt voters in swing states that Trump won who identify as “more moderate,” which aligns with Biden’s views.

But there are multiple strategies to beating Trump beyond Biden’s, Niebler said. And there’s a long campaign to consider. Public opinion can change rapidly based on debate performances and scrutiny on the trail, and as the economy fluctuates.

“I think many Democrats now think they have three or four, if not more, [candidates] who can beat Trump," Niebler said. “There’s a real danger of reading things 15 months out and saying, ‘The path to the White House for Democrats looks like this.’ Well, it might. But only if you’re this person.”

History also plays a role in which candidates are seen as electable, Niebler said. In a span of 45 presidents, the country has elected only one black man and no women.

“Electability arguments are problematic because we’re making assumptions about what our neighbors would do, when in reality we don’t know if they would or would not vote for someone based on identity markers," Niebler said. “We kind of assume they might not because we’re relying on old models of who has been elected in the past."