September 25, 2020: Updates on three new predictions

It is difficult keeping up with the numerous prediction. In the last couple of weeks, I've become aware of 3 additional models, which do not use polling data to provide forecasts. These are the Alan Abramowitz model, the Primary Model by Professor Helmut Norpoth and the Professor Graefe's Pollyvote Expert survey. In addition to these models, Dr. Allan Litchman on August 5, 2020 updated his 13 Keys to the White House.

I am grateful to Jared Farley for sending me the information on Abramowitz predictions and Allan Litchman updated forecast. He also let me know about updated forecast by Don Luskin. He is an Associate Professor of Political Science at New Mexico Military Institute in Roswell, NM.

(1) Alan I. Abramowitz is the Alben W. Barkley Professor of Political Science at Emory University and a senior columnist with Sabato’s Crystal Ball. His latest book, The Great Alignment: Race, Party Transformation, and the Rise of Donald Trump, was released in 2018 by Yale University Press. He used the data from 11 elections where an incumbent is running for re-election and approval ratings from late June.

EV-Trump = 289.6 + 4.68(NA), where NA = Disapproval - Approval rating for Trump.

The correlation coefficient = 0.64 and standard error = 91.4.

With an approximately 42% approval rating and a 57% disapproval, NA = -15, he predicted EV-Rep = 219, giving Biden a victory with 319 EV's. He also predicts Trump has a 29.5% chance of winning the election. Note with this equation, if Trump's net approval rate drops to 0, he would still be the predicted winner with EV = 290.

A second equation using late October approval ratings resulted in a better correlation = 0.92 as follows:

EV-Trump = 261 + 5.65(NA)

With this equation, if approvals equal disapprovals, Trump would lose with only 261 EV's, albeit an extremely close race. Abramowitz states the equation is based on approval ratings from the Gallup polls. To win 270 EV's, Trump would need an approval rating just 1.6% above his disapproval rating. Trump had several polls where his approval rating was higher than his disapproval ratings from February 16 to May 15, 2020.

The second equation is not supposed to be used until late October, but I feel I can cheat just a bit, since mail-in voting will be well underway by then. On September 13, Gallup shows NA = 42% - 56% = -14%, which results in EV-Trump = 182 and EV-Biden = 356. I prefer fivethirtyeight.com averaging of polls, which as of September 25, shows NA = -9.9%, EV-Trump = 202, and EV-Biden = 336.

See model details.

(2) Helmut Norpoth is a Professor of Political Science at Stony Brook University. Trump is the expected winner based on his model. He predicts Trump will win the election with 362 electoral votes and a 91% chance of winning. Professor Norpoth informed me that he directly calculates electoral votes from his model. I hope to learn more details. Certainly, if Trump wins, he will be well known as the forecaster who got it right. A link to his website is given at the end of this summary.

Dr. Norpoth indicated he will has prepared a paper on his method, most likely in late September to early October.

(3) Dr. Andreas Graefe is a Professor of Management at Macromedia University of Applied Sciences in Munich, Germany. The Polly Vote prediction combines forecasts within and across the following component methods to calculate its popular vote forecast: Polls (intentions) Expectations Models Naive forecasts. As of September 24, their composite forecasts indicates Biden will win the election with 326 electoral votes and a 70% chance of winning. The model also forecasts that Biden will win 52% of the vote.

One of the forecasts presented on the www.pollyvote.com website is the poll of experts. The survey consists of asking 12 university professors their predictions of the percent of the popular vote each candidate would receive. The latest survey had Biden with 55% of the popular vote. The surveys have been done each month. In September, the survey had Biden with 53.6% of the popular vote, so it appears his chances of winning are improving at least in the opinions of scholars.

I appreciate the work of these experts in sharing with us their predictions. This summary will be updated if the forecasts are revised.

David Lord

September 24, 2020

Links:

Alan Abramowitz, It’s the Pandemic, Stupid! A Simplified Model for Forecasting the 2020 Presidential Election

Helmut Norpoth Primary Model

PollyVote.com