I’ve been publishing projections on Twitter and Facebook for the last few days. This is the final one, using 538’s polling averages.
That looks pretty good for Biden. If you’re curious about the methodology, check out my projections from 2016.
- A Probabilistic Look at General Election Match-Ups: Clinton Vs. Trump & Sanders Vs. Trump
- A Probabilistic Look at the General Election Match-Up: Clinton Vs. Trump (21 June)
- A Probabilistic Look at the General Election Match-Up: Clinton Vs. Trump (27 June)
- A Probabilistic Look at the General Election Match-Up: Clinton Vs. Trump (6 Nov.)
That last one seems pretty far off. I’m grateful to Avery Yeates, who was one of my Summer Research Students this year. We tweaked the Maple program I had been using a bit. We separated out the congressional districts in Maine and Nebraska and we ramped up the variance in the individual state elections. The old variance made sense for a weighted coin toss, where the probability is static. It didn’t reflect the variance we see in election results. We’re estimating the probabilities that a voter will vote for a candidate. These change over time, in fact, the entire point of campaigning is to shift those probabilities. I need to look at correlating the states that move together for 2024.
With that said, here’s my final prediction.
My gut tells me that there will have been a big shift in Biden’s direction over the weekend. It’s looking more and more like he is going to win and people like voting for a winner.
In addition, the news has not been good for President Trump. We’re hitting records for the number of new cases of Coronavirus each day and some of the behavior we’re seeing from a small segment of Trump supporters is downright disturbing. If there are any swing voter’s left, I think that pushes them in Biden’s direction. Of course, I could be completely wrong. I’m least confident that Texas will turn blue. They had a huge number of early voters and so, fewer people to be swayed over the weekend. On the other hand, some of the outrageous behavior seems to be motivated by the belief that Trump could lose.
If you’re curious about my track record, this contains my prediction for 2018.
Meanwhile, here’s what I wrote in advance of the 2016 Election, in a lab I did with my Calculus students called “It’s Election Day 2016”
I’m bitterly disappointed that I couldn’t do this lab with my students this year.