The Very Unlikely Happens: How Donald Trump Might Win Again

Low probability events happen

While Donald Trump is in the worst position of any major party nominee since Bob Dole in 1996, he could win reelection.

Trump has a 7% chance to come back in the Economist model (shown below), and I’d put his odds somewhere between 5% and 10%.

Those low probability events happen though. And the failure to plan for them often leads to shock.

In November, I flew down to Austin and then made the 90 minute trek to Waco for the Baylor-Oklahoma football game. In the second quarter, Baylor took a 28-3 lead. The Bears had about a 93% chance to win; over the next two hours, it seemed like everything broke against the Bears. They couldn’t stop Oklahoma’s quick strikes. The Bears had a costly turnover, and they dropped an easy first down late. It took a lot to blow a 28-3 lead, but it happened. Despite having a 93% chance to win, my Bears lost.

I wasn’t prepared for the loss. I felt the 28-3 lead was basically a 100% chance of victory. The thrilling high of a 25 point lead against the Big 12’s best team made me treat that 7% chance of a comeback as zero. But my hubris didn’t actually shift the odds to zero; When Oklahoma came back, I was completely crushed. If I’d actually trusted the math that showed the Sooners could win, maybe I’d have felt a little less miserable.

An important way to avoid being ill-prepared for possible failure is by performing a premortem. Gary Klein explained the method in a 2007 edition of Harvard Business Review:

Research conducted in 1989 by Deborah J. Mitchell, of the Wharton School; Jay Russo, of Cornell; and Nancy Pennington, of the University of Colorado, found that prospective hindsight—imagining that an event has already occurred—increases the ability to correctly identify reasons for future outcomes by 30%. We have used prospective hindsight to devise a method called a premortem, which helps project teams identify risks at the outset.

A premortem is the hypothetical opposite of a postmortem. A postmortem in a medical setting allows health professionals and the family to learn what caused a patient’s death. Everyone benefits except, of course, the patient. A premortem in a business setting comes at the beginning of a project rather than the end, so that the project can be improved rather than autopsied. Unlike a typical critiquing session, in which project team members are asked what might go wrong, the premortem operates on the assumption that the “patient” has died, and so asks what did go wrong. The team members’ task is to generate plausible reasons for the project’s failure.

A premortem is also helpful because it can adjust our odds about Trump’s comeback. Maybe once we look at how Trump could win the election, we’ll think his odds are better than the 7% chance the Economist gives him.

For Trump to win, there are three scenarios. First, the polls or structural framing that he’s down so much could be wrong, which would mean Trump doesn’t face such a daunting comeback. Second, things could break Trump’s way. Finally, Biden could tank, and Trump—while still deeply unpopular in November—could win another election where enough people in crucial states view him as better than the alternative.

Maybe the polls are wrong, or the structural framing is wrong

The 2016 national polls didn’t miss by much—they had Clinton winning by 3 points and she won by 2.1 points—but they missed by 5.3 points in 538’s polling model in Wisconsin and more than three points in both Michigan and Pennsylvania.

There are three obvious ways the polls could be wrong, and then there’s always the potential for something we’re missing.

First, are the polls still missing whites without a college degree? Some 2016 polls missed because they didn’t weigh by education. And when educational attainment became a bigger dividing line in 2016, the polls were off. That mistake is less likely to happen in 2020 because many surveys now weigh by education. But maybe more of them vote this time, so even while they weigh by education, they aren’t capturing enough of them.

Second, the polls might suffer from “shy Trump voters.” Some theorize that social desirability bias might be in play where Trump supporters—knowing how loathsome he is to many—feel uncomfortable telling a pollster they’ll vote for Trump. This seems less likely to happen because even online surveys find consistent movement against Trump, and if you’re afraid of letting people know you support Trump, then why are you afraid to let a computer panel know? But maybe this impacts (I have joined the side that we should just use impact to avoid effect/affect, so sorry for making impact a verb if that bothers you) a close race.

Third, the polls might finally suffer from nonresponse bias. In the age of cell phones and spam calls, the response rate is sometimes as low as 1%. And there’s research that the people who respond—even if you get an accurate makeup of the electoral demographics by education—are somewhat different in-group. Often people who respond to polls are more likely to volunteer or be open to new experiences. So maybe we’ll hit a threshold where our sample is just too different within the groups. After all, how many people really just answer random unknown calls today? The counter is that 2018 polls held up well, and 2016 polling does well if you correctly shift the electoral breakdown.

Two other structural factors could matter. First, Trump or the GOP might succeed in suppressing the vote. Democrats have voted in record numbers in most races—even as they’ve had to wait in line far longer than they should—but maybe Trump can challenge mail-in voting, and if the virus is still a problem on November 3rd, Biden’s voters could disproportionately stay home. Second, the electoral college—which was 3 points more favorable to Trump than the popular vote in 2016—might lean his way even more in 2020. Trump could barely hold on in a bevy of states, including Texas, Arizona and Florida, while Biden runs up larger margins in California and finds getting closer in states Clinton lost costs him the presidency.

I tend to discount the above scenarios because I believe Trump’s 2016 win is best explained as Trump getting lucky against a historically unpopular opponent, but if you believe that interpretation is wrong, Trump’s odds are probably already better than 7%.

Things Break Trump’s way

There are four things that could break Trump’s way: the virus, the economy, a rally around the flag event or a Supreme Court open seat.

First, the virus might stop being an albatross. As the chart above shows, Trump’s approval rating and his approval rating of his handling of coronavirus are nearly the same. Maybe we get a vaccine announcement and voters reward Trump. Maybe some antiviral drug proves effective in reducing lethality. Maybe the new cases finally slow down with widespread masking, and this stops being a big issue by November. Heck, it’s such a weird era that maybe Trump succeeds in assigning blame to blue cities for the virus. Or maybe people just get tired and stop caring about the virus. The issue for Trump though is that even if these things happen, he needs his ratings to go back up after the virus gets dealt with. And it’s possible that a change in circumstances with the virus won’t mean people will forget some of the worst moments from his response to it.

Second, the economy could improve. Maybe Trump works out a deal for another round of stimulus payments, or maybe people really start spending money. After leading on the economy in hypothetical head-to-head polling in 2019 and early 2020, Trump’s now trailing Biden on that question in the most recent polls from Quinnipiac, ABC and Monmouth. But if he could retake the lead on that, maybe he could win.

Third, there could be a rally around the flag event. While I’d hate to have another awful event, maybe a terrorist attack happens and voters support Trump. Or maybe we get the dreaded second virus, and in a time of panic, people support the leader. There’s a good argument that in this much more polarized era any boost would quickly dissipate. But Trump had a three week run where the country rallied around him and approved of his handling of the coronavirus:

Finally, a Supreme Court vacancy could help Trump. If someone retires or dies, it could rally Republicans to support the party. Marsha Blackburn trailed in Tennessee throughout the summer of 2018, but around the Kavanaugh nomination, she took the lead. Any Supreme Court battle would be toxic, especially if Trump gets to replace a liberal justice after Republicans blocked Merrick Garland in 2016. If that happens, Republicans that dislike Trump’s demeanor and response to the virus might return to the party.

Biden goes into the tank:

There are a host of events that could sink Biden into being an unacceptable alternative to Trump.

Maybe the protests will continue and suburban white voters will backlash. That seems unlikely now as opinions on Trump’s handling of race relations are terrible, but he’s polling so terribly now maybe he can only go up:

Another possibility is that Trump’s campaign hammers Biden for his past views on race and criminal justice. Maybe that leads to apathy among non-white voters, and lower turnout allows Trump to eek out a win in the Great Lakes states.

Biden might also perform horribly in the debates, and maybe this “Biden is too old to be president” dog whistle from the Trump campaign takes hold. On day one, Biden will be the oldest person to ever occupy the oval office, so maybe this becomes an issue. Biden’s never debated someone like Trump, and with Trump down big in the polls, he might take a high-risk strategy that Biden blows.

There are two other Biden problems I could see. First, there could be another sexual assault allegation. The Tara Reade story took a while to gain media attention, but if someone else came forward, it could become a dominant story, and maybe that leads people to think, “both candidates are awful,” and tons of people vote third-party or stay home:

Another issue is what happens if Trump’s justice department claims Biden or the Obama Administration improperly investigated the Trump campaign. Bill Barr has been fiercely loyal to Trump, and this could become the dominant story in the waning days of the campaign. Maybe that shifts the election a few points, like the Comey letter did in 2016, and combined with a polling error, pushes Trump to another term.

Finally, maybe Trump convinces low-information voters—who could swing the election—that Biden is actually way too far to the left or represents some unpopular views. The above chart from Vox shows that Biden is slightly closer to the ideal candidate position on a seven point scale. But Trump is going to try and convince voters that Biden wants to defund the police, tear down statutes of George Washington and implement everything Bernie Sanders has ever supported. Those are lies, but maybe people believe that.

Trump is a huge underdog; he could still win

I haven’t assigned values to most of these things—the goal was to consider how Trump might win, not figure out the possibility all of them might happen in July or make this another 2,000 words—but my issue for Trump is that it’s probably necessary for his position to materially improve on the virus, and then he needs to find a way to have some other things break his way.

Trump’s been president for almost four years and his approval is underwater on the economy, coronavirus and race relations. His personal ratings are terrible too. Even if something breaks his way, the public is deeply skeptical of him. The public will not give him the leeway they’ve given earlier presidents, as shown by how quickly they soured on his handling of the virus and protests and evidenced by his tepid approval ratings throughout his presidency.

Trump could still win though. After looking at all those scenario, you might ascribe a lower probability than I do that Trump can come back. But it’s undoubtedly true that he has some chance. Even while Biden has the best odds of the 12 major party nominees this century, nothing is guaranteed. Sometimes your favorite team blows a 28-3 lead because the 7% event happens.

You can follow Kendall on twitter @kendallkaut. You can subscribe to the newsletter, which will publish at least twice a week on the election and politics at https://kendallkaut.substack.com