If you spend a lot of time on Facebook, you might see a post that says something like “We didn’t have seatbelts/safety features/basic concerns for our survival needs when I was a kid and we still survived” sooner or later.
If you haven’t seen them, think of the person who says “My grandma smoked every day of her life and she lived to 95” or “I used to drink 35 cans of Budweiser before driving and I never died” from Uncle Billy. These are all examples of “survivor bias”, which means this: these activities seem less risky than they are because you are one of the survivors. You don’t hear from people with similar stories, such as “I used to drink 35 cans of Budweiser before driving and I died right away, day one” from Uncle Billy (may he rest in peace), because Uncle Billy is dead.
Let’s leave the sad examples, and look at two that you might find on the Internet sometimes: The helmet and the plane. Wars are a good place to find survivors, so these stories both happen during a war.
The WWI Helmet
This one, which might not be true, goes like this: In WWI, generals started to worry when hospitals saw a lot of head injuries to soldiers in hospitals. Because of the increase, they thought that the problem might be that the new helmet – the Brodie helmet, specifically – might be hurting the head. Actually, the reason why they saw so many head injuries is that the helmets were letting more soldiers survive.
During World War II, the Americans wanted to lower the number of deaths in their aircrews. Many planes came back with bullet holes in three main parts: the body, the outer wings, and the tail. They had an idea to make the parts that had been shot by the enemy stronger. That sounds reasonable enough.
But, the reason why this is such a good example of survivor bias is that the only data they had was from the survivors. Before they could start to make the parts stronger, Abraham Wald, a Hungarian-Jewish statistician looked at the data and saw the mistake in their thinking.
Basically, the bullet holes in the body, outer wings, and tails on the planes that survived showed this: if planes could be shot a lot in those three parts and come back, that meant that being shot there was not very bad for the planes. If you guess (and you have to guess because you are not seeing all the evidence, as it’s somewhere in a war zone) that bullet holes are spread pretty evenly over planes when you look at all the planes that go out to fight, that means that the ones that didn’t come back probably had bullet holes in other parts. His solution was to think about the survivor bias, and suggest that they should make all the parts that didn’t have bullet holes stronger, to make survival higher.