Watching the US election over the last couple of days I have been struck time and time again how poorly polling and probabilaty predictions have been interpreted. With the arrival of Nate Silver’s FiveThirtyEight and the widespread adoption of quantitative poll aggregation techniques, it had appeared to me that the commentariat had learned how to read polling data. Yet, time and time again I was struck by how badly pundits on both British and US media interpreted the 90 and 97% probability of a Biden victory given by FiveThirtyEight and the New York Times. A 90% probability means only that 9 times out of 10 the election will fall one way. It makes no prediction on the type or strength of that victory. While it is clear that there has been institutional error in polling data, the misapplicaiton of probablity was what struck me the most. Wanted!
People that understand probablity to enter into journalism and political coverage to present clearly what a 90% probability means and how to interpret that. We here at Felix can’t stand another night of coverage like that.