Your election night crystal ball 📊 November 1, 2020
These numbers could help us call an early Biden victory
Thanks as ever for reading my weekly data-driven newsletter on politics, polling and the news. As always, I invite you to drop me a line (or just respond to this email). Please click/tap the ❤️ under the headline if you like what you’re reading; it’s our little trick to sway Substack’s curation algorithm in our favor. If you want more content, I publish subscriber-only posts 1-2x a week.
Well, here we are folks. After hundreds of hours of coding, writing, re-writing and coding some more I find myself staring down a mammoth election on the eve of polling day. Setting aside (for now) the fact that voters have been casting ballots for months, it is surreal for such a highly anticipated even to finally be upon us. Those of us who have worked tirelessly to cover the election and take part of it are on the brink of respite.
However, like any good story, there must be a climax before exposition. We must experience the election before we can really learn from it. And no good election-night watch party would be complete without the thrill of painstakingly tracking live returns as they trickle in. So I’ve taken the liberty of preparing a few things to keep your eye on as the results come in. Of course, you’ll also want to watch my Twitter feed, where I’ll post live analysis and real-time updates to our forecasting model based on predictions of final results in Florida, North Carolina, Georgia and Virginia.
I’m excited about the election. I hope our hard work will finally pay off.
Your election night crystal ball
These numbers could help us call an early Biden victory
Before getting to the data, here are two things to remember about making aggregate electoral college predictions based on partial returns. First, focus on county-level swings from 2016. The large number of them enable us to make high-powered predictions of what might happen across the county based on the demographic and political characteristics of those that have already returned. And second, states tend to move together, usually with a good amount of uniform national swing. Thus, if Biden is meeting his polls-based expectations in quickly-reporting states such as Florida and North Carolina, he’s very likely headed on a path to victory.
But now for the actual numbers.
At 6:00 eastern time, polls will close in parts of Indiana and Kentucky. By 6:30, we should receive initial counts of mail-in ballots and election-day results from them, with the first full returns coming in around 7:00 in the less populous rural counties.
These initial data will be a crucial time for our first modeling of the night. Matching up these early full results to a dataset of 2016 and 2012 county-level returns will allow us to predict trends in other areas of the country. If counties with more white, working-class voters have shifted by an average three or four points on margin toward Biden, we should begin thinking that this will be an early night with a swift Democratic victory. The same is true in suburban counties, which should shift to the left and also provide a lot more votes this year than in 2016.
In terms of specific counties to watch, I’ll be keeping my eyes on Boone, Fayette, and Campbell in Kentucky and Floyd, Harrison and Vigo in Indiana. If Biden is up ~2-3 points on Clinton’s vote margin there, the race is headed for a Biden victory. A 5 or 6 point swing gets him to landslide territory.
Next, at 7:00, the polls will close in Virginia and Florida.
Virginia is a great state for election data. The state board of elections publishes live precinct-by-precinct returns, including counts of absentee and election-day ballots. This offers even more statistical power to our comparisons of 2012 and 2016 electoral performance. I will be watching for an average statewide swing of about seven points from Clinton to Biden, which will affirm the prediction from the polls. Anything greater than a two or three point swing will also indicate a robust Biden victory nationally, though probably indicate a long night of counting.
By 7:30 we should also be getting initial tabulations, separately, of early/absentee votes and election-day votes in Florida. Florida is perhaps the best state for handicapping presidential elections as each county reports the total number of votes cast throughout Election Day, as well as the breakdown in vote method. This allows us to calculate the statewide percentages of ballots remaining that Trump will need to win the state, which we can match up against 2016 results to gauge how likely it is that he could win the necessary returns.
Here’s the number to watch: if Biden is doing about one and a half points on average better than Clinton did in fully-reporting precincts, he’s on path for a bare victory there. And since Florida is the key to Trump’s electoral strategy, any Biden victory there would crush his chances at re-election. The various election models I have built or played with stipulate that Trump’s overall odds of winning the election fall to 0.1% without it, as he’s also almost certainly losing the bluer northern battlegrounds in that scenario.
By 8:00 PM, we should know what type of night we’re in for. An early one, or a long count. But there are some states that can still provide a relatively early read on the race.
Polls will begin to close quickly across the battleground states. PA, MI, OH, NC, TX will all close at 8:00, with returns coming in in rural areas around half an hour later. We will repeat the same exercise as in Kentucky and Indiana, averaging the county-level swing toward Biden as final returns are announced.
At 9:00, polls in Wisconsin and Arizona will close. The former will not report many votes on election night, as roughly half will be cast via mail and unable to be counted until election day, which could take a while. Arizona, on the other hand, has been able to tabulate and verify early and absentee votes for weeks, which increases the speed of counting results. Returns will come fast at first in the rural areas, then slow as a mass of election-day ballots are counted in Maricopa County, home to Phoenix. If Biden can hold Trump to a few points on margin there, he will very likely win the state, and with it the election.
By 10:00 PM, we should turn again to Florida, where roughly 90% of ballots will have been counted (at least, that’s what has happened in the past). If I have not been able to call the election by then, the network decision desks will start revealing whether we’re in for a few days of counting absentee ballots in tight Midwestern states. But if Biden is headed for a easy victory, Florida will be called by roughly 9:30 or 10. Anything later will suggest a long night.
Unless the election is as close as it was in 2016 — which could happen, but probably won’t — we should know the probable winner of the election well before midnight. Maybe even around 8 or 9 PM.
That’s it. Cheers all. And good luck.
Posts for subscribers
October 31: 2020 looks like it will be less polarized than 2016. What’s the word for this? Dealignment? Re-realignment? Reverse polarization?
October 25: With nine days left, massive polling error is Donald Trump's best shot to win. He has run out of time to truly turn things around
October 25: If Biden beats the polls in Alaska, what should we predict in New Jersey? Thinking about a weird problem in 538’s election model
What I'm Reading and Working On
I wrote this piece last week. And this one. Also, here are two more. And a fifth. I’m so tired…
Thanks for reading!
Thanks for reading. I’ll be back in your inbox next Sunday. In the meantime, follow me online or reach out via email if you’d like to engage. I’d love to hear from you. If you want more content, I publish subscriber-only posts on Substack 1-3 times each week.
Alane sent me this picture of her cute tabby a few weeks ago.
For next week’s contest, send me a photo of your pet(s) to elliott[AT]gelliottmorris[DOT]com!