Culture et histoire

6 avril 2024 à 12:42

Survivorship Bias

Le post

During World War II, fighter planes often returned from battles bearing the scars of enemy fire. The Allies noted the sections that sustained the most damage and initially considered reinforcing these areas to enhance the planes' survivability. However, mathematician Abraham Wald proposed a counterintuitive approach. He suggested that the key might lie in the areas that weren't hit. The logic was simple yet profound: the lack of bullet holes in certain sections might mean that damage there was fatal, causing planes not to return. This insight shifted the focus to reinforcing the untouched areas. This historical anecdote underscores a vital principle in data analysis: sometimes, what we don't see is as important as what we do. It's a reminder that the absence of data can reveal critical insights, changing the way we understand and respond to problems. #PowerBI #DataAnalytics

Le débunk

The narrative holds truth, yet the term wasn't his doing. ✔ Abraham Wald, an Hungarian geometer and statistician, significantly impacted statistical decision theory during World War II. A University of Vienna geometry graduate in 1931, he escaped from Nazi Austria to the United States in 1938. His pioneering work on sequential analysis at Columbia University constituted a significant breakthrough in the field. ✔ Wald is considered the unofficial progenitor of "survivorship bias," drawing from his 1943 analyses on how aircraft survival correlated with incurred damage. This research informed aircraft optimisation during the Vietnam War, uncovering that areas on aircraft that sustained minimal damage were more prone to future failures, especially the engine. The concept of survivorship bias demonstrates a prevalent cognitive error where one focuses on what one believes in. We tend to prioritise visible success over unseen failures. This bias influences our interpretation of both historical and contemporary events. Stay skeptical and knowledgeable, (and follow C'est vrai ça ?) https://cvc.li/ZVBhy https://cvc.li/JgMnm https://cvc.li/JYLHG

Vous avez une question, une remarque ou une suggestion ? Contactez-nous en en indiquant le titre du débunk. Nous vous répondrons au plus vite !

Contactez-nous