We started a follow-up with data from the electoral race in the United States, before the Democratic convention and once the candidates for president and vice president of the two parties are known
Democrat Joe Biden is the favorite of the polls, the experts and the betting, but just think of 2016 to remember a truism: being the favorite does not ensure you win a presidential election.
The favorites also move. Donald Trump started 2020 with history on his side. He was a president seeking reelection with the economy growing, which has traditionally been a favorable situation. But that advantage was wiped out with a pandemic that is killing 1,000 Americans a day and has triggered an economic crisis with unpredictable consequences. Trump stayed ahead in the forecasts until March, when the virus began its expansion and the Democratic primaries opted for Biden, who was possibly the rival he feared the most. Four months later, the Democratic candidate leads the polls with six or seven points over Trump.
Biden's margin is considerable, but not final. First, because there are three months until the elections (things happen). Second, because a vote is uncertain even on the last day (there are surprises). And third, because 2020 is not a normal year. There are no gurus or algorithms that really know who is going to win, but neither do we have to go to the extreme of believing that the elections are absolutely unpredictable. The balance is in thinking in probabilities.
And in that sense, what do we know about these elections? There are at least five useful forecast sources that are updated almost every day.
The prediction I trust the most says that Biden has a 72% chance of winning and Trump a 28% chance. That's the forecast from FiveThirtyEight's survey-based mathematical model , which statistician Nate Silver has produced for a decade. A similar model from the weekly The Economist , which in addition to polls gives great weight to economic variables, raises Biden's options to 88%. Instead, bettors give Biden 60% and Trump 40%, leaving the vote almost a toss-up.
Finally, this year the Good Judgment company is publishing the predictions of its experts, the “super forecasters” of the investigations of Philip Tetlock and Barbara Mellers, who have been raising the Democratic options from 40% in February to 79% today. It is the other forecast that I value the most.
Let's say Biden has a 75% chance of winning and Trump a 25% chance. What does that mean exactly? Trump's surprise victory in 2016 sparked a debate about how to best communicate the odds. A debate I think exaggerated. It's true that at a glance, Biden's advantage looks great: He's three times as likely! But one sees immediately that it is not so bad. Trump has a choice of four to win, which is more or less the same as Leo Messi has to miss a penalty with Barcelona. Sometimes it happens and the fans know it. People understand probabilities when they matter to us. And, of course, if our lives are in it: that's why nobody plays Russian roulette with a loaded revolver saying that 16% seems little to them.
It is also a fallacy to say that percentages are useless because "anything can happen." It's like saying that it doesn't matter if the penalty is thrown by Messi or if you throw it, that you also put it in. It is also not irrelevant to give Trump 25% or just 1% choice, even if Biden is your favorite. How useful would it have been to know the night before Brexit that the UK could actually leave the EU? May the future surprise you less than the majority is worth money.
As of now, Trump's race is against the clock. It's what these numbers say. If November 3 rolls around and the polls still give Biden six or seven points, his options will be slim. According to the FiveThirtyEight model, with that advantage and a vote that was tomorrow, the probability that the polls will be wrong and Trump will win is only 10%.
Trump needs things to happen. And of course they can happen. It is easy to think of a universe where the president announces a vaccine, goes back two or three points, the race loses interest because Biden is still a favorite, young people demobilize, the epidemic brings participation down to historical lows and Trump is re-elected (although he loses the popular vote). But nothing special can also happen, or the opposite can happen: that Biden shows up in the debates, the economy worsens and Trump is devastated. A posteriori it will be easy to tell the story of these elections. It is in our nature, as the psychologist Amos Tversky said: “People predict very little and explain everything”. We predict by making up stories, he added. It is inevitable to do so, but you can try an experiment: instead of choosing your favorite narrative, which seems most likely to happen between now and November, try to think of half a dozen scenarios. Suddenly the future looks more uncertain, blurred. Of all those possible universes, you ask yourself, which one will be mine?