Sam Trenholme's webpage
Support this website

More predictions


November 2 2020

Some more predictions for tomorrow’s election.

==The approval ratings prediction==

One way to see how likely an incumbent president will get re-elected is to look at their approval rating. Historically:

  • In 2012, Obama had a 52% approval rating and got 51% of the vote.
  • In 2004, Bush had 48% approval and 51% of the vote.
  • In 1996, Clinton had 54% approval and 49% of the vote.
  • In 1992, Bush had 34% approval and 37% of the vote.
  • In 1984, Reagan had 58% approval and 59% of the vote.
  • In 1980, Carter had 37% approval and 41% of the vote.

Trump currently has a 44% approval, which means he’s going to, at best got 48% of the vote. Then again, Bush beat Gore in 2000 with just under 48% of the vote, and Trump beat Clinton in 2016 with 46% of the vote. According to Nate Silver, a 2% popular vote victory for Biden means he only has a 41% chance of winning, a 3% popular vote victory gives him a 67% chance of winning, and a 4% popular vote victory gives Biden an 89% chance of winning.

So, while the approval ratings indicate that Biden will get a popular vote victory, it does not mean he will get the electoral vote victory he needs to win, but, based on 2004 and 1992 (the last two times a Republican incumbent has been up to reelection), it gives Biden about a 67% of winning.

Here in 2020, with things as polarized as they are—I did not see street picketing for candidates in previous elections like I have this year—there is a chance that Trump will energize enough voters to give him a vote closer to his approval rating. If Trump gets only 44% of the popular vote, Biden has a 97% chance of winning.

==Polly Pollster the AI==

Here’s another prediction: An AI, called “Polly Pollster”, has been observing social media and seeing how people vote based on social media postings. While I have not been able to verify it, Polly has supposedly predicted Clinton’s victory in 2016, some elections in Canada, and Bernie Sanders’s winning of Iowa in the primaries.

Her prediction is the same prediction I made yesterday, but with the states I marked “purple” (could go either way) blue: Biden gets 367 EVs compared to the 171 EVs it says Trump will get.

==The Needle in 2020==

This year, The New York Times will only have a “needle” for three states: Florida, Georgia, and North Carolina. In order to translate how results from these states affect the election results, here is what 538 predicts based on how those three states go:

  • All three go to Trump: 50% chance Biden victory, 48% chance Trump victory, 2% chance tie.
  • FL, GA Trump NC Biden: 8% chance Trump, 92% chance Biden
  • FL, NC Trump, GA Biden: 22% Trump, 76% Biden, 2% tie
  • FL Trump, GA and NC Biden: 3% Trump, 97% Biden
  • FL Biden, GA and NC Trump: 3% Trump, 97% Biden
  • FL and NC Biden, GA Trump: < 1% Trump, > 99% Biden
  • FL and GA Biden, NC Trump: 1% Trump, 99% Biden
  • FL, GA, and NC Biden: < 1% Trump, > 99% Biden
  • FL Biden, GA/NC unknown: < 1% Trump, > 99% Biden
  • FL Trump, GA/NC unknown: 32% Trump, 66% Biden, 2% tie
  • GA Biden, FL/NC unknown: < 1% Trump, > 99% Biden
  • GA Trump, FL/NC unknown: 23% Trump, 76% Biden
  • NC Biden, FL/GA unknown: < 1% Trump, > 99% Biden
  • NC Trump, FL/GA unknown: 28% Trump, 71% Biden, 1% tie
  • NC and GA Trump, FL unknown: 33% Trump, 65% Biden, 2% tie

Point being, with these three states, we will be able to predict a Biden victory, but not a Trump victory, because Biden may still have the Rust Belt firewall.

Comments for blog entries can be seen on the forum.