American Research Group, Inc.

 

2020 US Presidential Election Pollster Ratings

 

Updated: February 3, 2021

The following table lists the preliminary accuracy of pollsters in the November 3, 2020 US presidential election using the measure of polling accuracy proposed by Martin, Traugott, and Kennedy.

An accuracy measure of 0.00 indicates that the odds ratio of the survey results matches the odds ratio of the actual results (the closer to 0.00, the better).

 

2020 US President
Biden Trump Odds
Ratio
Accuracy
The Hill/HarrisX 49% 45% 0.99 -0.01 
IBD/TIPP 50% 46% 0.99 -0.01 
Emerson 50% 45% 1.01 0.01 
Reuters/Ipsos 52% 45% 1.06 0.05 
JTN/RMG 51% 44% 1.06 0.06 
Rasmussen Reports 48% 47% 0.93 -0.07 
SurveyUSA 52% 44% 1.08 0.08 
Fox News 52% 44% 1.08 0.08 
USA Today/Suffolk 52% 44% 1.08 0.08 
NYT/Siena 50% 41% 1.11 0.11 
Economist/YouGov 53% 43% 1.13 0.12*
CNBC/Change 52% 42% 1.13 0.12*
NBC News/WSJ 52% 42% 1.13 0.12 
USC Dornsife 54% 43% 1.15 0.14*
Quinnipiac 50% 39% 1.17 0.16*
CNN 54% 42% 1.18 0.16*
Actual 51.31% 46.86%

 

*The accuracy measures for the Economist/YouGov poll (sample size of 1,363), the CNBC/Change poll (sample size of 1,880), the USC Dornsife poll (sample size of 5,423), the Quinnipiac poll (sample size of 1,516), and the CNN poll (sample size of  886) fall outside the 95% confidence intervals. This represents approximately 44% of all final national polls. 

The accuracy measures for the other 11 polls fall within the 95% confidence intervals.

The average absolute accuracy for all polls is 0.085.

The average absolute accuracy for the 11 polls where the accuracy measures fall within the respective 95% confidence intervals is 0.061.

The average absolute accuracy for all polls in the 2016 presidential election was 0.070. The average absolute accuracy for all polls in the 2012 presidential election was 0.055. The average absolute accuracy for all polls in the 2008 presidential election was 0.037.

2020 State ratings:

Florida

Ohio

Michigan

Wisconsin

Iowa

North Carolina

Arizona

ARG Home