This is my follow-up article about the presidential election polls and predictions about the election results. You can read the first article here.
The best tool we have to determine the outcome of elections is polls. Each poll has its own bias and we can get rid of some of the bias in individual polls by combining the results of each individual poll in one giant poll. Fivethirtyeight is a website that does this for us. Its method isn’t fool proof though. We can get rid of some of the bias, but not all of it.
Right before the 2016 presidential election, on November 7th, Fivethirtyeight estimated that Hillary Clinton had a 67% chance of winning the election. Yet, Donald Trump crushed Hillary. There are several reasons why this happened. I am not going to waste your time to discuss these. The most important factor was that voters didn’t reveal their true intentions when asked by pollsters. As a result, polls failed to predict the outcome of the 2016 elections. Take a look at this screenshot of Fivethirtyeight’s website from November 7, 2016 (the day before the 2016 election):
Today, Fivethirtyeight thinks Biden has an 87% chance of winning the election. An almost slam dunk case. I disagree.
I disagree for two main reasons. First, the polls are wrong. Second, recent polls are even more biased because Trump contracted COVID-19. Four years ago just around the same time Trump was in the middle of the “Billy Bush tape scandal” and experienced a similar deterioration in polls. Let’s remember what was said in the media at the time:
“Five days after America learned what Donald Trump likes to (non-consensually) grab women by, voters continue to withdraw their consent for his candidacy.
New polls show Trump trailing by 9 points nationally, by that same margin in Ohio, and tied with Hillary Clinton in Utah. What’s more, the L.A. Times/USC tracking poll, whose aberrant voter model has put Trump ahead for most of the campaign and all of the past month, now has him tied with the Democratic nominee. The Ohio poll, conducted by the Baldwin Wallace Community Research Institute, finds Clinton leading Trump 43 to 34 percent in a four-way race, and 48 to 38 in a head-to-head.”
Can you take a moment and try to remember how you felt about the election results around that time? Trump, supposedly, was 10 points BEHIND Hillary in Ohio and tied in Utah? Just three weeks after the publication of that article Trump destroyed Hillary by 8 points in Ohio and 18 points in Utah. Can things change this dramatically in 3 weeks? Not probable. Clearly the poll results around October 12, 2016 were extremely biased and FAR FROM predicting the outcome of the 2016 presidential elections. I believe recent poll results that lead 538 predict a sweeping Biden victory are biased in a similar fashion. Before going state by state, let me give one example that will also make you doubt a landslide Biden victory.
Florida will probably determine the outcome of this presidential election. Right now fivethirtyeight estimates that Joe Biden is ahead by 3.4 points. You will notice that Biden opened the margin after Trump’s diagnosis at the beginning of October and it is starting to narrow.
Now take a look at the results of recent Florida polls below. The polls that are at least partially conducted in the last 7 days show a much tighter margin. There are a total of 4 polls during the last 7 days and Joe Biden’s average margin in these 4 polls is only 0.5 points. Please keep in mind that these polls are still BIASED. Four years ago fivethirtyeight predicted that Trump would win Florida by 0.1 points, but Trump won Florida by 1.2 points. If we assume that fivethirtyeight’s current BIAS is also 1.1 points in favor of Biden, we can see that Trump will win Florida by 0.6 points if elections were held today (last 4 polls’ average is 0.5 points for Biden minus 1.1 points for Trump will give us 0.6 point Trump victory). That’s why I currently believe that Trump will win Florida, not Biden.
In this article I am going to assume that the current polls in each state have the same bias as they had in 2016 and I will adjust the current poll results to estimate the true intentions of the voters. At the end of this article, we will have a better idea about who will win the presidency.
1. Alabama
Fivethirtyeight 2016 prediction: Trump wins by 21.6 points
Actual result: Trump won by 27.7 points
Fivethirtyeight 2020 prediction: Trump wins by 19.2 points
Insider Monkey Prediction: Trump wins by 25.3 points
Trump: 9 – Biden:0
2. Alaska
Fivethirtyeight 2016 prediction: Trump wins by 7.3 points
Actual result: Trump won by 14.7 points
Fivethirtyeight 2020 prediction: Trump wins by 6.8 points
Insider Monkey Prediction: Trump wins by 14.2 points
Trump: 12 – Biden:0
3. Arizona**
Fivethirtyeight 2016 prediction: Trump wins by 3.3 points
Actual result: Trump won by 3.5 points
Fivethirtyeight 2020 prediction: Trump loses by 3.1 points
Insider Monkey Prediction: Trump wins by 0.0 points
Trump: 23 – Biden:0
4. Arkansas
Fivethirtyeight 2016 prediction: Trump wins by 19.8 points
Actual result: Trump won by 26.9 points
Fivethirtyeight 2020 prediction: Trump wins by 18.4 points
Insider Monkey Prediction: Trump wins by 25.5 points
Trump: 29 – Biden:0
5. California
Fivethirtyeight 2016 prediction: Trump loses by 22.5 points
Actual result: Trump lost by 30.1 points
Fivethirtyeight 2020 prediction: Trump loses by 31 points
Insider Monkey Prediction: Trump loses by 38.6 points
Trump: 29 – Biden:55
6. Colorado
Fivethirtyeight 2016 prediction: Trump loses by 3.4 points
Actual result: Trump lost by 4.9 points
Fivethirtyeight 2020 prediction: Trump loses by 11.4 points
Insider Monkey Prediction: Trump loses by 12.9 points
Trump: 29 – Biden:64
7. Connecticut
Fivethirtyeight 2016 prediction: Trump loses by 11.1 points
Actual result: Trump lost by 13.6 points
Fivethirtyeight 2020 prediction: Trump loses by 22.9 points
Insider Monkey Prediction: Trump loses by 25.4 points
Trump: 29 – Biden:71
8. Delaware
Fivethirtyeight 2016 prediction: Trump loses by 12.2 points
Actual result: Trump lost by 11.4 points
Fivethirtyeight 2020 prediction: Trump loses by 26.4 points
Insider Monkey Prediction: Trump loses by 25.6 points
Trump: 29 – Biden:74
9. District of Columbia
Fivethirtyeight 2016 prediction: Trump loses by 70.8 points
Actual result: Trump lost by 86.8 points
Fivethirtyeight 2020 prediction: Trump loses by 83.4 points
Insider Monkey Prediction: Trump loses by 88 points
Trump: 29 – Biden:77
10. Florida**
Fivethirtyeight 2016 prediction: Trump wins by 0.1 points
Actual result: Trump won by 1.2 points
Fivethirtyeight 2020 prediction: Trump loses by 3.4 points
Insider Monkey Prediction: Trump wins by 0.6 points (we explained above how we estimated this margin)
Trump: 58 – Biden:77
