A lot of superstitious Democrats may look at the unveiling of FiveThirtyEight’s initial presidential election forecast for November and experience a sense of dizziness and nausea: It gives Trump a 29 percent chance of reelection, precisely what FiveThirtyEight gave him in its final forecast of the 2016 election.
So for all the talk of Biden’s very steady polling lead, does this mean Trump is as likely to win as he was on Election Day of 2016?
Not exactly, as Nate Silver explains:
In 2016, the reason Trump had a pretty decent chance in our final forecast was mostly just because the polls were fairly close (despite the media narrative to the contrary), close enough that even a modest-sized polling error in the right group of states could be enough to give Trump a victory in the Electoral College.
The uncertainty in our current 2020 forecast, conversely, stems mostly from the fact that there’s still a long way to go until the election. Take what happens if we lie to our model and tell it that the election is going to be held today. It spits out that Biden has a 93 percent chance of winning.
So it’s mostly the unknown shape of the future that boosts Trump’s chances for a comeback from his current standing against Biden. Time, and not much else, is currently on the incumbent’s side.
Silver goes through a lot of factors that influence his forecast, and you should go read his entire explanation. But I’d summarize the uncertainty factor about November by borrowing Donald Rumsfeld’s famous distinction between known knowns, known unknowns, and unknown unknowns.
One “known known” is that presidential elections usually tighten as Election Day nears, a tendency that should be strengthened by today’s powerful forces of partisan and ideological polarization (though polarization also reduces the number of swing voters, which works against comebacks by a candidate at Trump’s level of disadvantage). Silver points to a reminder of leads in past presidential contests at this point in the cycle:
Four of these elections showed incumbents gaining ground in the popular vote between mid-August and Election Day: Bush ’04 (4.9 percent), Reagan ’84 (2.2 percent), Carter ’80 (12.4 percent), and Ford ’76 (24.5 percent). Reagan ’84 was on his way to a landslide win in both August and November. Similarly in 1980 Reagan had a big cushion to work with and Carter was bound to make some gains because of the inevitable decline in support for independent candidate John Anderson. And in 1976 it wasn’t surprising that the candidate of the party that had won 49 states four years earlier wasn’t going to lose by 20-plus points to a nearly unknown Democrat whose religion and regional traits alienated many liberals. The incumbent comeback most relevant to Trump’s situation is probably Bush ’04. But it’s worth noting that W. was the rare incumbent who succeeded in making his reelection a “choice” election as much as a “referendum” thanks to attacks on Kerry’s war record and “flip-flops.” Trump is obviously trying to do that, but so far hasn’t laid a glove on Joe Biden.
Another known known that Silver touches on is Trump’s (and his party’s) Electoral College advantage, which in 2016 meant that the Republican could win 57 percent of the electoral votes with only 46 percent of the popular vote (2.1 percent less than Hillary Clinton). Looking at battleground polls, he suggests it could very well happen again:
That reflects the fact that the tipping-point state — the state that would provide the decisive 270th electoral vote — is somewhat to the right of the national popular vote. More specifically, our projection as of Tuesday had Biden winning the popular vote by 6.3 percentage points nationally, but winning the tipping-point state, Wisconsin, by a smaller margin, 4.5 percentage points …
[A]s a rough rule-of-thumb, perhaps you can subtract 2 points from Biden’s current lead in national polls to get a sense for what his standing in the tipping point states looks like.
I would personally add three more “known knowns” to the equation, and neither is good for Trump: The minor-party vote is almost certain to be down this year, and there will not be a lot of overconfident would-be Democrats casting “protest votes” or deliberately staying home. The magnitude of these changes from 2016 isn’t clear, to be sure, but it’s hard to deny they could matter.
And finally, a long-term trend toward early voting (in person and by mail) will almost certainly intensify thanks to COVID-19 fears, no matter how often Trump fulminates against it. That means the candidate trailing may have marginally less time to make up a deficit before already-cast votes become baked into the cake.
So let’s move on to the “known unknowns.” The biggies are COVID-19, the economy, and the presidential debates.
FiveThirtyEight tried to build COVID-19 trends into its forecast, but Silver admits it mostly just adds to the uncertainty surrounding the direction of the race. It’s obviously good for Trump if infections and deaths go down or if a vaccine looks more imminent. But how good is another question: Even as voters didn’t seem to blame Trump for the arrival of the pandemic, they may not give him much credit for its departure if it doesn’t seem he had much to do with it.
Similarly economic trends have been so wild this year that it’s unclear how to get a fix on them as a factor influencing the election. Silver writes:
One model based on second quarter GDP projects Trump to win -453 (negative 453!) electoral votes, for example. But if you built a model based on third-quarter GDP, which is expected to be highly positive, it might predict a Trump landslide.
This isn’t to say that we don’t employ a fundamentals forecast of our own. We do, but it’s much less confident than others, and it receives relatively little weight in the overall forecast.
Trump seems to think (like many candidates trailing in the polls) that debates might turn things around, but there’s not much evidence debates change votes in presidential elections that much, and in any event, either candidate is as likely as the other to commit the kind of gaffe that really could affect the contest. Again, it’s a known unknown but probably not a game-changer.
A last “known unknown” that could really be a big deal is voter turnout:
We also expect turnout to be harder to predict this year based on primary elections held during the pandemic that had highly variable turnout — which, in turn, could lead to more polling error.
Aside from COVID-19 fears, both parties are engaged in an unprecedented legal and legislative battle across a vast landscape to gain an advantage via electoral rules that make it easier (Democrats) or harder (Republicans) to vote. And that’s in addition to the traditional get-out-the-vote activities of both parties, which can typically only be measured after the fact.
So what about “unknown unknowns,” the things that come out of nowhere and change everything? By definition, of course, there’s little we can do even to speculate about such possible developments as a war, a major terrorist attack, a major outbreak of domestic violence, the death or disability of one of the candidates, or even a second pandemic. Typically, big shocking events tend to produce a “rally-around-the-flag” effect for incumbent presidents, though perhaps not for long, and perhaps less infallibly for a president who has aroused such strong and fixed views about himself.
One possible development that is on the borderline between a “known unknown” and an “unknown unknown” is the possibility that Trump will try to contest and nullify an election defeat, or at least throw it into doubt at the cost of massive public unrest. As someone who has warned of this possibility for months, I still do not see a clear path from there to a re-inaugurated Trump next January.
In the end, Nate Silver’s basic argument that uncertainty over future events is the main reason to question Joe Biden’s polling lead is sound, so long as you take into account that uncertainty cuts both ways:
There’s the chance that Trump could come back — but there’s also the chance that things could get really out of hand for him. Our model thinks there’s a 19 percent chance that Biden will win Alaska, for example, and a 13 percent chance that he will win South Carolina. The model also gives Biden a 30 percent chance of a double-digit win in the popular vote, which would be the first time that happened since 1984.
But when you look at all the factors that might save Trump’s bacon, it’s striking how many of them are simply out of his control. If Biden’s lead slowly dissipates over time, Team Trump will mostly be condemned to watch it and hope it speeds up.