A classic example (Google survivorship bias plane for context)

There's a very popular image of an outline of a plane with red dots that indicate where it was hit by enemy fire. The image from the Allied forces in WWII was a data compilation of where planes were hit after they had returned back to base.

A statistician Abraham Wald, rightfully pointed out that reinforcement against fire should be made to the parts showing least damage because those planes did not make it back to base to be recorded! The map of where planes were hit and had made it safely back to base, showed that these planes could return even after sustaining damage in those areas.

An example for start-ups/ business

Even in business, we tend to look at start-ups that made it with a similar business model or idea, and use that as justification for doing the same. With the qualifier: We will do better by avoiding their mistakes.

The whole time, there is an incredible amount of data left unused of all the non-survivor start-ups. The ones that did not make it represent a data set that is left unexplored because of survivorship bias.

One last example

Ever wondered how old buildings are all beautiful? They were not!

Only the beautiful, well-built ones survived.

The ugly or poorly constructed ones were torn down for new structures!

Next time, I'll be thinking about both sides of the story before making a judgment or using a dataset as justification for actions.