Polling published over the last several weeks has former Vice President Joe Biden out in front of President Donald Trump by as much as 12 points. The average seems to be about 10 points. In 2016 the margin of the Clinton victory was supposed to be 30 points and higher. Almost every polling firm and pollster agreed with that assessment, except one. Now he says the president has a 91 percent chance of being reelected in 2020.

Mediaite reported on Wednesday that Stony Brook University professor Helmut Norpoth is doubling down on his “Primary Model.” That process, which uses primaries to predict turnout, has correctly predicted five out of the past six elections since 1996 and every single election but two in the past 108 years.

“The Primary Model gives Trump a 91 percent chance of winning in November,” Norpoth said. “This model gets it right for 25 of the 27 elections since 1912, when primaries were introduced.” The model botched the 1960 election of John F. Kennedy and the 2000 election of George W. Bush. Both easily explained. Kennedy stole that election by cheating in Illinois and Texas and Bush barely squeaked by in 2000 in the Electoral College and lost the popular vote.

According to Professor Norpoth, Biden is in a bad position because of his poor showing in the first primary races: “The terrain of presidential contests is littered with nominees who saw a poll lead in the spring turn to dust in the fall. The list is long and discouraging for early frontrunners. Beginning with Thomas Dewey in 1948, it spans such notables as Richard Nixon in 1960, Jimmy Carter in 1980, Michael Dukakis in 1988, George H.W. Bush in 1992, and John Kerry in 2004, to cite just the most spectacular cases.”

The professor has a valid historical point. Ed Muskie in 1972, Scoop Jackson in 1976, Ted Kennedy in 1980, Gary Hart in 1984, Bob Dole in 1988, Paul Tsongas in 1992, and others since were all early frontrunners who stumbled. All underperformed in the primaries. None even made it to the convention.

In politics, as perception is reality, the press and popular culture can make or break candidates on a whim. One must only recall the campaign result of Howard Dean to prove that proposition.

Norpoth’s model, which predicted Trump’s victory nine months before the 2016 election, predicts that the president will win over Joe Biden by a wider margin in the Electoral College (with 362 electoral votes versus the 304) than he earned against Hillary Clinton.

That would be roughly akin to the shellacking Barack Obama gave John McCain in the 2008 presidential election.

If Norpoth is again proved correct, even by not that great a margin, then one would think the very basic premise of modern political survey research would be called into question.

Who do you think would win the Presidency?

By completing the poll, you agree to receive emails from LifeZette, occasional offers from our partners and that you've read and agree to our privacy policy and legal statement.

However, given most polling is expressly manufactured to support a campaign’s effort for momentum and not as a legitimate snapshot of public attitudes, campaigns will continue to pay media outlets big money for rigged results and politicians will continue to use the results to proclaim the inevitability of their victories.