Watching Election night — a data driven approach
I follow politics more than most and have studied the data pretty carefully. I am not a data scientist, but I don’t suck for a guy with just a business degree.
I decided to write this final 2020 election blog to share how I am watching tomorrow night and what I am looking for.
I am mostly watching online
I am following Nate Silver’s fivethirtyeight.com live blog and I will be watching Nate Cohn’s New York Times section Upshot including watching “The Needle” on FLA, GA and NC . I will have the TV on watching PBS, but will be watching CNN and Fox on other live streams with the volume down.
Here is what I am looking for:
The most likely tipping point state will not be decided on election night.
It is almost impossible for Trump to win without Pennsylvania. At the same time, Biden goes from a 90% favorite to an underdog if he loses PA. Biden leads by ~4.9% in fivethirtyeight polling average. Enough for him to sustain a 2016 size polling error, but not by much.
PA will not be a factor on Tuesday night. Early ballots are not counted quickly and full results won’t likely be available until Friday or Saturday. Don’t pay much attention to partial results in PA. PA, like most states, will not report mail in ballots by county and thus it will be statistically difficult to predict who has won unless it is a total blow out.
There are only 7 battleground states that matter. Trump must win all 7.
There is no reason to tune into CNN to hear, “The Decision Desk has called New York for Biden! or The Decision Desk calls West Virginia for Trump”. We know who is going to win those states.
There are seven states (large enough and close enough) to tip the total electoral college in one candidate’s favor or another. Here is my summary of them:
I have already pointed out that PA will not be called on Election night, so we need to watch the other 6 states. They will tell us if we have a winner on election night. If Biden wins any of FLA,GA,NC, OH, TX, AZ, then he will almost certainly have won. If Trump sweeps those states, then he is the favorite and we wait for PA.
A small asterisk. If Biden wins AZ and not any of the other states, then he will only be tracking for 269 EC votes (a tie). If that is the case we need to watch a few smaller contests.
Iowa, NE2 (Omaha) and ME2 might matter
Iowa. Trump is up 1.5% in the polling average here. It’s a battleground state, but it’s only 6 EC votes. If Biden only wins AZ out of the above 7 states then its feasible Iowa matters.
Also, single EC votes in Nebraska2 or Maine 2 could put Biden at 270. NE2 will be called on Election night. ME2 is unlikely to be called, they have rank choice voting and that takes some time.
Determining winners by state will be much harder this year.
Because of the pandemic, there are a ton more mail-in votes than ever before. Most states will not report county level results by vote type. So it is going to be very hard to create models that project winners based on partial vote counts. We are going to have to wait until much larger amounts of the vote in each state are counted to declare winners
I would expect state election calls about 3–4 hours after polls close, but maybe its 5–6 hours if the election gets close.
Watch the NY Times Needle in FLA, GA, NC
These three states report county votes by vote method- mail in or in person. This will allow Nate Cohen and team to do their data science magic and the needle will be our most likely early indicator of what is happening. But data feeds to the needle are sometimes a little inconsistent. So while I am hopeful that this will be great resource, it might not be. Some of the major networks might have decent county level projections, but I would trust Nate Cohen most.
What about Michigan, Minnesota, Wisconsin
Biden is comfortably ahead in these states. If Trump wins those states then the polling error might be 2 times 2016 and then Trump is winning bigly. Further those states count slow, so we are not likely getting election night results from them that will matter on election night.
Hour By Hour Guide
7PM-9PM Eastern
Watch the Needle and read fivethirtyeight.com live blog. The Needle will tell us if we have a winner. If Biden is an early projected winner of FLA or GA, then it is over early. NC has a lower chance of being an early indicator because it counts votes that arrive after election day, so it won’t be as clear a signal as FLA, GA.
9–11 PM Eastern
We should start to get decision calls on the eastern states based on almost full vote counts. FLA,GA,OH are the only states that really matter. Again NC will be harder to call.
Also polls will close in Texas and Arizona and results will start trickling in.
11PM — 3AM Eastern
Texas results should be getting very clear. Trump must win there to have any chance. And Arizona is likely to be decisive state if the race is still alive.
Also, we should have a chance to see what results look like in NE2 and Iowa in case they factor in.
3 AM Eastern
Probably time to go to bed. If Biden has won any of the key battleground states, then he is almost certainly the next President. If Trump has swept those battleground states, then we will have to wait for PA. But Trump will be the favorite to win PA if he sweeps those other states.
Senate Races
This is also critically important. But it’s a lot less likely we know who controls the Senate on Election night. Here is my math on that.
Democrats need +3 to win Senate control. Republicans will almost certainly be+1 in Alabama and the Democrats +1 in Colorado. Democrats next best chances are AZ, NC and Maine. Maine won’t likely be called on election night because of rank choice voting and given NC counts votes past November 3rd it might not be called. Georgia Senate races are likely headed to January runoffs. So the only likely way Dems get to +3 is AZ and less probable wins in IA and MT.
