It has been an interesting couple of days, hasn’t it! My youngest daughter Kaitlyn and I have exchanged more texts in these last two days than we have in the rest of our lives combined! Like me, she is a political junkie, except maybe more so. Both of us are also into data… and there has been a lot of data flying about the last three days.
I make my living as a data analyst. When I am tuned into a topic I can make some very good predictions. I had some pretty strong thoughts about where the American Election might be headed, so I thought it would be interesting to put my thoughts in public before the polls closed.
Here is the tool that I used for the technical part of my determinations. The author is Éric Grenier, and I have been a fan of his for many years.
The tool has a cool feature in that you can use a slider to see the effect of a deviation from the latest polling numbers. The data has been updated ever so slightly from when I used it but what you see now is pretty close to what I used.
Here was my thinking. Donald Trump exceeded his polling numbers by 2.2% in 2016. The reasons the pollsters gave was that it was because they didn’t properly apportion a poll share to white uneducated males. They told us they had it fixed for 2020. The reason I am hearing this time round is the “secret Trump supporter”.
I thought it was simpler than either of those excuses that had been given pre and post the election.
1. Conservative voters always exclude their poll numbers. This is largely because older voters, who tend more conservative, have a better turnout on election night than younger voters.
2. Like 2016, Democrats did not have a candidate that they were enthused about.
My thinking went: If Trump could beat his polling by 2.2% in 2016, then he could likely beat it by 2.0 this year, and I adjusted the slider accordingly.
The state that really stood out as the swing state was Pennsylvania. At a 2.5 percent change in the slider, Trump would pick up Nevada, but still lose the election. At 3.0% was were Pennsylvania flipped into the Trump camp.
In my mind then Pennsylvania was key to winning the election this year, and so I made my first prediction:
1. Whoever wins Pennsylvania will win the election.
Note: as I write this, Pennsylvania is still in the Trump camp, but is trending strongly towards Biden. By sometime this morning it will be in the Biden camp, and he will win in by about 120,000 votes.
I didn’t think Trump would beat his polls by 3% and so I made my second prediction.
2. Biden will win Pennsylvania
At the time of writing this post, Biden was behind by 18,000 votes with 275,000 to count. He was winning 80% of the mail in ballots.
Based on my above analysis, I thought the election would be close, a lot of states would be close, and there would be recounts and lawsuits flying! So I felt pretty safe in making my third prediction:
3. The ultimate winner will not be decided for over a week.
I also knew that the mail in ballots would be a huge factor in this election, especially in how they were being counted after the fact in several key states. This would have the effect of Trump initially leading, and then losing ground as the mail in ballots were counted. And so I made my fourth prediction:
4. Trump will be ahead as of 11:00 p.m. (E.S.T.)
In fact he was! At 11:00 p.m. he was elected or leading in 278 electoral college ballots. By the time I went to bed that night Trump was elected or leading in 296! I must say I was tempted to second guess myself.
But I still thought that my original analysis was correct. And that was that if Trump couldn’t take Pennsylvania, some other states might be a toss up, but he would be maxing out at about 259 electoral college votes. That is why I had made my original fifth prediction:
5. Trump maxes out at 259 (or less) out of 270 electoral college votes.
So that is how I made my predictions.
What kind of outcome were you expecting? Have you been on a roller coaster of emotions over the last three days?
As usual your thoughts and comments are welcome. My Facebook friends managed to keep things civil despite their varied backgrounds. I hope we can do the same.