The debate tonight is likely to be a doozy. It is one of the few chances left for the candidate and campaign that is likely behind by ten points nationally and six or seven points in the tipping point states to introduce variance into the race. Variance is the only hope that they have as the trend is working against them. Variance could lead to a bigger loss, or it could lead to the possibility of getting the race to the margin of litigation.
The election is ongoing. Another couple of million people voted today. The outcome on November 3rd has less variance today than it did yesterday and significantly less variance than it did on October 15th.
Fact: more people have now early voted in North Carolina (2,4317,317) than the number of people who voted for Donald Trump in North Carolina (2,362,631) in 2016.
— Dave Wasserman (@Redistrict) October 22, 2020
All 2,417,317 people who have already voted in North Carolina are now indifferent to new information. Their votes won’t change if the President is a calm, empathetic individual with a firm grasp of actuarial implications of moving individuals to parallels markets. Their votes won’t change if the President has to be injected with Thorazine while on the debate stage. Their votes will not change. Some of those 2.4 million votes will be for the President. Some of those 2.4 million votes will be for the challenger who is in a commanding national position. These votes are locked in.
Tomorrow, even more votes will be locked in.
And on Saturday and Sunday, even more votes will be locked in.
Each day moving forward reduces the universe of people who can change their mind.
Let’s do a simple thought experiment on the importance of variance and locked-in votes. The Economist has Vice President Biden projected to be leading in North Carolina 51-49 on October 22. Let us say that there are 5,000,000 major candidate voters in the state so the state winner is the first one to 2,500,001 votes. Let us assume that the projection is perfectly right, so the Vice President is currently supported by 2,550,000 voters and the current President is supported by 2,450,000.
If we had a single election period where everyone goes to the polls on Tuesday November 3rd, the President needs variance to move 1%+1 votes to win the state. 1% +1 is not a ton of noise. Maybe the Vice President showed too much compassion for his son, or he ordered a Cheese Steak with Swiss instead of Provolone or horror of horrors, he indicated a preference for mustard based barbecue. If we assume variance is symmetrical, there is an equal chance that the Vice President wins by 3% instead of 2% but from a direct electoral perspective, an additional vote above the minimum needed coalition is not particularly useful.
Now let us assume that we have a 2 election period and that the distribution of voters between periods is random. the distribution of preferences on 10/22 is the same (51 B 49 T). By the end of Thursday, October 22, half the votes will be in. These individuals can’t change their vote. The President still needs to make up 50,001 votes. However those votes need to come from the pool of individuals who are potentially amenable to new information. This means there needs to be a 2% swing from the pool of voters still out there. This is harder to achieve than a 1% swing. There might be some voters who have already voted whose opinions have changed but their votes are constant.
If we make assumptions about the composition of the first period of early voters, the numbers will change. If we think that the early voters are more likely to be Biden voters, even more variance is needed as the pool of potential Biden-Trump switchers in the election day pool is smaller than it would be if voting in the early period is not correlated with preferences. If we think that the early period is heavy on Trump voters, then less variance is needed.
North Carolina is just a good teaching case as the numbers are round enough and the margins are close enough. Variance at this point is the best hope for the candidate who is running behind beyond normal polling errors in the tipping point states. And as early voting continues, the amount of variance needed has to increase because the number of people who are potentially information sensitive is continually declining.
So tonight, expect WTF as the President has minimal downside to a fiasco which either does not change the contours of the race or increases Vice President Biden’s lead. He has no institutional loyalty to the Republican Party. He has only upside of perhaps doing something that could get current soft Biden voters and undecided voters to flow to him. So expect variance.