Hillary Clinton will be the next president of the United States. How many times was it said, in the fall of 2016, that the race for the White House was decided? The inconceivable victory of Donald Trump it dislodged pollsters, journalists and political analysts, even Trump’s own advisers.
Four years later the polls again bet against the republican, while he cries out already seen . Can history repeat itself? If they were wrong then why shouldn’t they do it again, as Trump claims?
There are fewer undecided and now polls take education into account, but experts ask for caution
In defense of pollsIt must be said that in 2016 they nailed the popular vote. The day before the election, they gave Clinton a 3.2 point lead. He won by 2.1.
The “historic failure,” in the words of Courtney Kennedy, director of survey research at the Pew Research Center, was in state polls. They are the ones that matter in the presidential elections. Trump lost the popular vote but won the electoral college.
It was a hell of an election: two historically unpopular candidates, a different winner in the popular vote and in the electoral college (it has only happened in five of the 58 presidential elections in history) or Trump taking three states (Pennsylvania, Michigan and Wisconsin, with his 46 electoral votes) by less than one point.
These presidential ones seem easier for the forecasts. There are fewer undecided. In 2016, a record number of voters (more than 6%, double the number in 2012) arrived on Election Day without knowing who to vote for, and a disproportionate number of them were decided at the last minute by Trump. This year, two months before the appointment, there was only just over 1% undecided.
Bugs have also been fixed. The polls have made adjustments so that each state’s sample is representative in terms of education, an indicator that proved decisive four years ago, says Kennedy, who led the committee of experts that examined the mistakes made. Voters with higher education were overrepresented, fictitiously inflating Clinton’s margin.
“In the US you didn’t have to worry if you had too many whites in the sample with or without university studies, they voted the same. But in 2016 there was a huge rift. Non-college whites overwhelmingly voted for Trump, while college students voted slightly for Clinton, “explains Sean Trende, senior election analyst at Real Clear Politics (RCP). It occurred with particular intensity in the hinge states (they range from party to party) in the Midwest, which polls painted Democratic blue but ended up leaning toward Trump.
Kennedy, very critical of the explosion of polls in recent years and their quality, cautions that many pollsters have corrected the bias in education, but not all, and that it is not a magic bullet anyway.
Trende, too, is skeptical: In the 2018 midterm elections, the corrected polls failed again, predicting Democratic governors in Ohio, Florida, Iowa, or Democratic senators in Missouri, Florida and Indiana. “I’m worried that maybe something is slipping away,” he muses. His hypothesis is that it is not only about educational level, but also about where voters live: urban and suburban areas versus rural areas. “If you include more whites without studies but they are university students or waiters, you have not solved the problem, what you need is to include farmers,” he reflects.
“There are more reasons to be optimistic about the polls but there are still many reasons to be cautious. Nobody should look at them and conclude that the race is over, “says Kennedy. A final turnaround is less likely than in 2016 but can never be ruled out, he warns.
The expert sees participation as an element of uncertainty. Predicting who will vote, something that was decisive in 2016, is always a challenge. This year, with the pandemic and a record early voting by mail, even more. Among other things, by the ability of novice voters to cast valid votes or states to process them.
Biden has it better than Clinton, but the stars may realign with Trump, Trende warns. In RCP’s median polls, the Democrat leads by 8.9 points nationally, up from 5.5 for Clinton at this point. “The question is whether it will happen like in 2016, when the margin narrowed in the last stretch. Biden achieved a 10-point lead in the summer with our second wave of Covid-19 and the Black Lives Matter protests, then down. Now, with the contagion of Trump, it has grown again ”.
His calculation is that the president could win the electoral college even losing the popular vote by 4 points. It is a margin not seen since 1876. Of course, Trump would be the first president to be elected twice without winning the popular vote.
It is also necessary to measure whether Biden’s advantage in key states is wide enough to guarantee victory if the polls were again as wrong as in 2016. Trende is also not clear: “The polls give him an advantage very similar to the one Clinton had at this point. It’s disturbing how similar it is. “
A lesson from four years ago is not to trust yourself. “In the last few weeks, polls were stopped in states like Michigan or Wisconsin, because people assumed that Clinton had it won. They couldn’t identify the narrowing of the last moment that happened, “says Trende.
He points to a third element that worries him: the approval of Trump’s discharge. “It is one of the mysteries of this election: right now, even with everything that is happening, it is 45%. Presidents typically get their approval rating plus one point in elections, so Trump should be 46% in polls. It has 42%. So there is a sizable chunk of voters who say they approve of your work but will not vote for you. Maybe they are people who at the time of voting remember that, in reality, they like what they do ”.
Does Joe Biden have a chance of defeating Trump?
Total votes: 0