In 2016, one poll gave Nate Silver of FiveThirtyEight enough angst to write a 2,000-word takedown, pitying and humoring the “outlier.”
The USC/L.A. Times tracking poll saw what Silver and the soothsayers missed: a submarine of silent Trump support.
“I’m tired of hearing about the poll from Donald Trump fans such as Reince Priebus, Matt Drudge and Donald Trump himself,” said Silver. “They frequently cherry-pick that poll because it consistently shows much better results for Trump than the other surveys.”
When Donald Trump “shocked” the world, the L.A. Times poll proved one of the few to see it coming.
In fairness, Silver gave Trump a better chance of winning than most in his milieu.
This year’s incarnation, the USC Dornsife Daybreak poll, mimes the rest of the big media offerings—Biden has led by double-digits since the start. It’s over.
But not everyone at USC Dornsife believes that. Two new questions added to the Daybreak poll—reported separately—find the popular vote much closer than suggested. They also predict President Trump will win the Electoral College again.
The Daybreak poll, like the big media polls, asks voters who they’re voting for—the “voter intention” question.
Yet, researchers are asking two additional questions they say could “harvest the wisdom of crowds” and unearth the mystical “shy vote.”
The “social-circle” question asks voters what percentage of their friends, colleagues and relatives are voting for either Donald Trump or Joe Biden. The “state-winner” question asks who they think will win their state.
The team used this method for the 2016 U.S. presidential election, the 2017 French presidential election, the 2017 Dutch parliamentary election, the 2018 Swedish parliamentary election, and the 2018 U.S. midterm election. In all five, the social-circle question was more accurate than the traditional “Who are you voting for?”
In 2016, the social-circle question accurately predicted the winner of each state—Donald Trump’s Electoral College victory.
Researchers say this question works because it widens and diversifies the sample size and allows voters to be honest with their choice.
So, what do researchers find so far?
“The social-circle question is predicting Biden will win the popular vote, but by a much slimmer margin than what’s being predicted by the standard voter intention question—in the USC Dornsife Daybreak Poll and most others.”
And what about the Electoral College?
“When we calculate how many electoral votes each candidate could get based on state-level averages of the own-intention and social-circle questions, it’s looking like an Electoral College loss for Biden.”
The researchers said the social-circle question in 2016 was more successful than both the voter-intention question and aggregate polls in predicting the winner of Michigan, North Carolina, Pennsylvania, and Wisconsin—all of which went to Trump four years ago. Aggregate polling didn’t get a single one right.
So, what happens when they ask voters who will win their state?
“This state-winner question produces even more pessimistic Electoral College results for Biden,” researchers said.
This would make logical sense. Asking who will win one’s state offers even more anonymity than asking that of one’s social circle. In an age when two-thirds of Americans reportedly conceal their political opinions, the value of anonymity is obvious.
The pollsters at USC give no specific numbers and say they’ll release their findings closer to Tuesday’s election.
Perhaps, they’re mindful of what happened last time. In 2016, the L.A. Times tracking poll did such a good job of recording reality that those uncomfortable with the results demanded they disfigure reality. That disfigurement is perhaps evident in this year’s Daybreak incarnation.
But not all national pollsters chime in with the double-digit Biden cadre, as 2016’s most accurate pollsters show the presidential race at a dead heat, and President Trump taking the lead in key states.
At the state level, The Trafalgar group on Thursday said President Trump has stretched leads in Florida (three points) and in Michigan. In Nevada, Trafalgar says Biden leads by just two points.
Trafalgar also has Trump with a narrow lead in Pennsylvania.
Trafalgar is credited with the “neighbor question” the polling firm says unearths a shy vote, much like the social-circle and state-winner questions USC researchers say points to an Electoral College win for President Trump.
In 2018, Trafalgar was alone in predicting Ron DeSantis beating Andrew Gillum in Florida’s gubernatorial race.
At the national level, Rasmussen finds 30 percent of black Americans are voting for President Trump—yes, that’s three-zero—capping rapper Lil’ Wayne’s endorsement this week.
Curiously, the same poll finds 43 percent of nonwhite voters (excluding black voters) are voting for President Trump. That number is down slightly from the week’s 45 percent average.
That daily tracking poll on Friday found a three-point race between the president and Biden, with Barack Obama’s former vice president leading 49 percent to 46 percent with a 2.5 percent margin of error.
On Thursday, Biden held a 48 percent to 47 percent lead, identical to Trump’s lead on Wednesday.
Biden was 12 points ahead, according to that same metric four weeks ago.
Rasmussen also said the president’s approval rating hit 51 percent—two points higher than President Obama on the same day in 2012—marking 10 days of approval figures above 50 percent.
That number hovers between Gallup’s 46 percent on October 1, and The Hill/Harris poll’s 49 percent on Thursday.
At the national level, the Investor’s Business Daily (IBD)/TechnoMetrica Institute of Politics and Policy (TIPP) presidential election tracking poll on Friday found five points between Donald Trump and Joe Biden, yet warned the race appears much closer in the crucial swing states, with both candidates locked at 48 percent apiece.
IBD also found Biden garnering just 78 percent of the black vote, while President Trump takes 50 percent of the Hispanic vote to Biden’s 48 percent. Trump leads by a point among seniors.
What is clear: someone is very wrong about all this. And mainstream pollsters won’t survive another 2016-style Hey Rube.
Numbers! We got numbers! We got stacks and stacks and stacks of numbers! (with apologies to the late Perry…) Figures don’t lie but liars figure! (not Mr Clemens, you can look it up; I did…) And especially: There are lies, damned lies, and statistics! (origin too murky for even comic relief…)
It sure would be nice if there existed serious professional self-policing amongst serious professionals. If serious professionals (begging the question of whether such a sub-species exists) worried one tenth as much about cleansing bias (aka, “conclusion first, analysis (if any) later”) as they did about occupational birth control, we would be better for it.
Thank you Mr Gage, for the revealing glimpse of the sausage-making that is political polling — another oxymoron of our age. Saves the rest of us the legwork.
My neighbors are overwhelmingly for Trump.
Trump is set to carry every state he did in 2016 and several more. This election he is going to add Nevada, New Mexico, Minnessotta, and New Hampshire. The Republicans are also on track to retake the House and gain in the Senate. The fake news is in for a knockout blow. Go out and vote!