As requested by several members, we have updated our previous stats report of US Election polls. We highlight some potential statistical pitfalls of the (aggregate) polls with states such as Texas, Louisiana, Missouri, and Indiana at most risk of surprise outcome. Modeling and methods of panel construction are beyond the scope of discussion in this brief report.
The figure below shows the states we consider at highest risk of having out of consensus outcomes. Our analysis considers: 1) the total number of public polls conducted for each state in 2020, 2) the population size, and 3) the candidate's lead according to the latest polls.
States with highest risk of a surprise outcome
Risk score
Black represents states with no public polling data found in 2020
Despite Texas State having over 39 qualifying public polls to-date in 2020, its relatively large population size and tight race make this a new battleground to watch in the days ahead. We have also highlighted states like Georgia (GA), Missouri (MO), Ohio (OH), Illinois (IL), Indiana (IN), and Tennessee (TN) as having highly uncertain outcomes. As we will show below, states such as GA, MO, IN, and TN (Rep) have far too few public polls conducted in 2020.
In the figures below we break down our analysis with illustrations for each factor we took into consideration.
Latest polls by State
Blue – Biden points advantage Red – Trump points advantage
Total number of polls by State
Black represents states with no public polling data found in 2020
Populations per (public) poll conducted in 2020
Million population per poll in 2020
Black represents states with no public polling data found in 2020
In conclusion, the relatively large number of state polls conducted in 2020 for battleground states such as Wisconsin (WI) has left other traditionally red/blue states at much higher risk of surprise outcomes – overwhelmingly towards the GOP.
