I was in Washington DC this time four years ago - a week before the 2016 election. The mood was eerie, not in the least because of all the morbid Halloween decorations. With skeletons hanging from trees, carved pumpkins on porches and fake gravestones littering front yards, the suburban vistas felt strangely like a scene from a zombie movie.
The nightmarish feeling wasn't confined to the Halloween celebrations. FBI Director James Comey had just announced the "reopening" of an investigation into Hillary Clinton's private email server, with new material surfacing on the seized laptop of disgraced former Democrat politician Anthony Weiner. The Clinton-Kane campaign was in damage control. Trump was ahead in the polls in Florida, and performing well in other swing states. The uncertainty of the Comey investigation, furthermore, was hurting Clinton's national popularity ratings. The febrile mood led some to cautiously predict a Trump upset.
Fast-forward to now, and the scene is very different. American today feels more like Camus's The Plague. Over 220,000 Americans have died so far in the coronavirus pandemic, and daily case numbers are rising in a majority of states. Four weeks ago, Trump himself was in hospital recovering from an unexpected bout of Covid-19. The prospect of a vaccine by November - a Trump campaign promise - is increasingly unlikely.
Coronavirus has seriously hurt Trump's re-election prospects. Despite a recent narrowing of the polls, many pundits think Trump has little possibility of re-election. The Economist 's election forecast says that Trump has a mere 1 in 20 chance of winning the electoral college; Nat Silver's FiveThirtyEight website gives Trump just a 12 percent chance of victory. NBC Political commentator Dave Wasserman, who predicted a Trump victory in 2016, says that the 2020 presidential race is fundamentally different, and that most of the fundamentals favour Biden. So, obviously, a repeat upset is unlikely.
But, but, but … is this really the case? The reliance on polling data may be leading pundits to overlook some important features of the election. The "shy Tory" effect - which pertains to people who vote for conservative parties while evading pollsters - may be particularly pronounced in this election, where Trumpism is more taboo than ever.
It's instructive, therefore, to also look at other indicators.
First, no incumbent with Trump's very high level of GOP primary support has ever lost his re-election - Trump is very popular with actual Republicans. They even turned out for Trump in a primary where he ran unopposed. Many Trump voters will march through a storm to vote for him. (I mean this literally, as Hurricane Zeta is due to hit the Gulf of Mexico in the next few days.)
Second, Trump will probably improve his vote amongst minorities. He will do well with Hispanics, which will matter in the swing state of Florida. He is popular among Asian voters. Believe it or not, Trump will better his performance with African-Americans - particularly men. Trump lost among African-Americans by about 82 percentage points in 2016 but has closed the gap in support to about 71 points this year, according to figures in FiveThirtyEight last week.
Third, Trump's support for manufacturing jobs and Biden's ambitious green energy policies, will give the Republicans a bump in the Rust Belt states. In the third debate, for example, Trump managed to draw Biden out on his commitment to transition from an oil industry to a reliance on renewables. A Democrat-lean poll of Pennsylvania voters published after the debate put Trump two points ahead of Biden (48 percent to 46 percent).
Finally, Trump's early voting and vote by mail numbers are looking healthy in the swing states of Florida, North Carolina, Michigan and Wisconsin. More registered Republicans than Democrats have cast their ballots in Michigan and Wisconsin, for example, despite the fact that Democrats were predicted to vote in far greater numbers prior to election day.
The final vote in these states will be close, but it is possible that Trump will win all four. If he wins Pennsylvania, it's curtains for Biden.
As Cary Grant said in The Awful Truth , "You're wrong about things being different because they're not the same". In many respects, Trump is better placed to win now as an incumbent. He's not seen as risky as he was in 2016.
Biden, on the other hand, may be seen as a liability by many voters - in part because of Trump's attempts to paint him as a puppet of the far left and as an enemy of the mining and resource sectors. This, rather than Trump's frenetic style of governance, may be the risk that many voters have in mind as they head to the polls.
Don't get me wrong. Biden still has a good chance of winning. But November 3 could still go either way. Pundits tipping a Biden landslide may get a surprise.