top of page

One Week to Midnight: What Will We Know, and When Will We Know It?

So why is this election night different from all other election nights?

Yes, the Toteboard is intentionally invoking the trope for the traditional Jewish Passover celebration, which remembers the miseries inflicted upon a nation by a malevolent tyrant, celebrates a hard-earned liberation from oppression, and expresses hope for promises yet to be fulfilled. But historical and theological significance aside, this election season really has been like no other, and the vote tabulation next Tuesday fittingly threatens to degenerate into chaos and confusion. In the shadow of a pandemic, precincts everywhere are (or will be) overtaxed by an unprecedented number of mail-in and absentee ballots, while republican saboteurs labor to intimidate voters, reject ballots, and undermine the credibility of the whole process through a sordid campaign of disinformation and litigation. And just in case this isn’t crazy-making enough, several key states employ arcane laws that both restrict how soon ballots can be counted and provide flexibility for how late they can be received (in some cases, up to two weeks after election day!), opening up the distinct possibility that we may not see the full results and may not actually know the winner for quite some time. Might as well end the Year From Hell with the Election From Hell, right?

But fortunately, while we may not “know” who won, we may have a pretty good idea by the end of the evening, or even sooner, though this requires looking to something other than the conventional methods for making projections. Exit polls will be especially useless in a year when so many people vote by mail, and because some states probably won’t be called on election night, it really won’t work to sit up and wait until one candidate’s shares add up to 270 or more electoral votes. So, what’s the alternative? Interestingly enough, the answer lies in closely watching three different three-state clusters, and extrapolating from there. The starting point (which may change somewhat by next Tuesday) is that Biden begins with 21 reliable states (counting DC) for a total of 233 electoral votes, while Trump begins with 20 red states totaling 163 electoral votes, which leaves a total of 142 EV’s up for grabs (that includes 1 vote for a tossup district in Maine). As you can see, when you need only 37 of the remaining 142, the math looks pretty good for you. And here’s how to tell how good the math looks:

The Firewall Triad: The cleanest and easiest way for Biden to put this one away is to snag those three rust belt states (MI, PA, WI) that Trump won narrowly four years ago, and the polls currently show Biden leading all three by about 5 to 8 percentage points, with Pennsylvania looking like the closest and Michigan looking like the safest. These leads are better than “narrow” but probably not quite what you could call “comfortable,” as the numbers look only a little bit better than they did for Clinton, which of course conjures up the specter of another system-wide polling error. Still, the pollsters in this cycle have supposedly learned to adjust for voter education level, and there doesn’t seem to be the potential for the same glut of late-breaking undecideds that went for Trump in 2016, so there’s reason to be more optimistic about polling accuracy this time around. With a combined bounty of 46 EV’s, this Triad would bring Biden’s total to 279, effectively ending the night early and leaving us free to howl at the moon, one and all. But through some accumulation of bad karma or just a nasty coincidence, these are among the states that are least likely to have complete results on Tuesday night. Officials and poll workers in all three states are not permitted to begin counting absentee votes before Tuesday, and in Pennsylvania and Wisconsin, they can’t even start processing absentee ballots until then (Michigan can do so a day earlier, which isn’t much help). To make things psychologically more distressing, there’s a good chance that these particular states will be reporting the in-person votes first, possibly not counting late-arriving absentee ballots for days (that’s still being litigated), so the early numbers could look disproportionately Trumpy, which might set the stage for republican cries of a “rigged” election if and when the numbers turn around later on. In short, there are a lot of forces conspiring against getting a good read on the Firewall Triad states right away, and we shouldn’t hold our breaths or empty our Xanax bottles waiting for them to come through.

The Insurance Triad: Counter-intuitively, the states that may give us a clearer sense of how Biden is doing are a cluster of three states (AZ, FL, NC), two burgundies and an aspirational purple, where the polls show Biden leading by a narrower 1 to 3 percentage points, and where the results could be known sooner rather than later, thanks to laws that allow for more efficient ballot processing and counting. If Biden gets an early-ish call for even one of these states, the way votes tend to correlate across states make him a more substantial frontrunner than he is already, i.e., going from something like a 7-1 favorite to a near shoo-in. This could rightfully give rise to a huge collective sigh of relief on Team Joe, but the Toteboard is obligated to offer a few bits of caution. First, it looks like the order of tabulation may be reversed in these states, where the mail-in and early in-person votes could be the first to be counted, which may give Biden the appearance of a more secure lead than he really has. Second, it should be remembered that the calculus employed here is based not on Biden actually hitting 270 electoral votes Tuesday night – that’s a heavy lift when the Firewall states will probably still be dicking around with their returns – but on statistical likelihoods, and some people are a bit uncertain when meandering around this kind of statistical analysis. For example, assessing Biden’s odds of winning the election after he carries, say, Florida, is not the same as assessing his odds of winning after carrying Florida but also losing Arizona and North Carolina, as the latter amounts to what numbers geeks call a “conditional probability” assessment. That is, if Biden carries Florida, that automatically suggests his odds have just improved in the other two states. If he suddenly loses them, well, then the numbers have to recalibrate. You can actually see how this works on Nate Silver’s interactive Exploration page, and feel free to contact the Toteboard if you need a primer or if noodling around seems to produce strange results. So, in the scenarios we just discussed, if Biden wins one of the three Insurance Triad states, he turns into an overwhelming favorite, i.e., a 98% favorite if he wins Arizona, a better than 99% favorite if he carries Florida or North Carolina. But, if he carries only one of the three states, and Trump carries the other two, then it gets a little more complicated. If he carries only Florida, then he’s a 10-1 favorite (which is still better than where he is now). If he carries only North Carolina, then he’s an even stronger 13-1 favorite. But if he carries only Arizona, which may not be as tightly correlated with other states, then Biden is only a little better than a 4-1 favorite, which is still pretty damn good, and better than where Hillary was, but probably not good enough to guarantee a good night’s sleep. To summarize, if Biden wins even one of the states from this Triad, he either remains on track to win or becomes an overwhelming favorite, depending on which state he wins. If he wins two of them (or three), then it really is very, very close to being over.

The Landslide Triad: Part of what makes Biden such a strong favorite at this point in the horserace is that he has a little bit of breathing room, that is, he has several possible routes to the promised land. And when you have a national lead of about 9 percentage points, states that looked fairly red even a month ago are starting to look like true toss-ups today. The polls indicate that Iowa, Ohio, and (gasp!) Georgia are all basically dead-heats, with Ohio showing an ever-so-slight tilt toward Trump. It’s not clear if these states can end up being early reporters – Iowa had that monumental meltdown after the democratic caucuses, Georgia is doing everything it can to wrestle the crown from Florida for election mismanagement, and Ohio has an idiosyncratic mechanism for vote collecting and reporting – but if it’s clear on election night that Biden has snagged even one of these, you can be pretty confident of a landslide. Yes, it seems kind of strange to say that it’s over if Biden wins Iowa, with its 6 measly electoral votes, but that’s what makes numbers so wonderful. And just in case you’re wondering what happens if Biden loses every state in the Landslide Triad and every state in the Insurance Triad, well, that means he’s probably underperforming his polls nationally by a good 4 or 5 percentage points, and it means we’re kind of right back where we started, i.e., needing to ride the Firewall Triad across the finish line. He might even be a slight underdog with that scenario, but let’s all just keep thinking good thoughts and hoping it doesn’t come to that.

And like they say, next week in Jerusalem.


bottom of page