Survivor bias is a logical error or fallacy where you concentrate your analysis on people or things that made it past some selection process as opposed to everyone involved, usually because of a lack of visibility or availability. This almost always leads to a false conclusion and the tendency to show causality when there really is none. If you're wondering about the picture above it shows a map of damage to WW2 planes coming back from missions, statistician Abraham Wald took survivor bias into account when trying to figure out how to minimize bomber losses to enemy fire. Most people would look at the picture and conclude that the places where there are red spots are where reinforcements should be added, of course the exact opposite is true because those planes hit in areas outside the red spots didn't actually make it back!
I heard a radio presenter today that said "it's really good news that all the people calling his show who have the new Covid variant say that it's mild" - can you spot the fallacy?
No comments:
Post a Comment