Polling firm USC Dornsife's Daybreak Poll added two new questions to its survey: What percentage of your social contacts do you expect to vote for each of the presidential candidates? and, Whom do you think will win the election in your state?
Without those two questions, their poll shows Joe Biden leading President Donald Trump by 11 points. With them, the gap is much closer. And when it comes to the Electoral College, Trump beats Biden.
Dornsife's pollsters tested those questions in five elections - the 2016 U.S. presidential election, the 2018 U.S. midterms, the 2017 French presidential election, the 2017 Dutch Parliamentary election, and the 2018 Swedish Parliamentary election. In all five, "The social circle question predicted election outcomes better than traditional questions about voters' own intentions," according to the pollsters.
And responses to the state-winner question confirmed the results.
Jim Key, director of marketing and media relations for USC Dornsife, said those two questions also accurately predicted state-level results better than the regular "own intention" question. "In fact," wrote Key, "data from the social-circle question in 2016 accurately predicted which candidate won each state, so it predicted Trump's Electoral College victory."
Why? "We believe there are three main reasons. By asking people how their social contacts will vote, we're implicitly gaining a larger and more diverse sample of participants. Second, it can be less embarrassing or intimidating for someone to tell a pollster that their friends plan to vote for an unpopular candidate than to report their own intention to vote for that candidate."
And, thirdly, wrote Key: "We are all influenced by our social contacts. Asking about the voting intention of someone's social circle could offer a glimpse [into] that person's future beliefs, sort of like a crystal ball."
The state-winner question, wrote Key, "produces even more pessimistic Electoral College results for Biden."
This is the same result predicted by Stony Brook political science professor Helmut Norpoth using his Primary Model. It's based on voter behavior during the primaries and has been tested back to 1912, the first year that primaries were instituted.
In the 27 elections since then, Norpoth's Primary Model has accurately predicted the election outcome in 25 of them, exceptions being the 1960 election (Kennedy vs. Nixon) and in 2000 (Gore vs. Bush).
For 2020, Norpoth's model has Trump beating Biden by a greater margin than he beat Hillary in 2016. Norpoth told Fox News:
The key to the November election is the primaries. The early primaries are giving us a lot of information. Based on that, Donald Trump won them very easily in his party.
Joe Biden had a great deal of trouble, pulled it together, but on balance it is that stronger performance of primaries that gives Donald Trump the edge in November.
People have forgotten how Joe Biden did in New Hampshire. He was terrible. He got 8.4 percent of the vote, which is unbelievable for a candidate with any aspirations of being president.
Norpoth's model gives Trump a 91-percent chance of beating Biden on Tuesday, capturing 362 Electoral College votes to Biden's 176.
It's all a matter of asking the right questions and looking at the right data.