How Statistics Saved The War

All too often, people focus efforts in the wrong direction. In this new world of big data plenty of companies analyze their competition and the marketplace to make strategic decisions. This can be a good method if you're well versed in statistics and reading between the lines. Unfortunately, people tend to make a lot of assumptions without realizing it and these assumption can often lead to negative results.

A great way to describe this actually lies within a World War II story. During the war a group of mathematicians was assembled to solve many problems the allies were facing. This group, called the Statistical Research Group, or SRG, was similar to the gathering of great minds in the Manhattan Project. The difference was in the problems they were trying to solve. With such a gathering of brilliant minds, the military came to the SRG asking for recommendations on where to increase armor on their planes. They were provided statistics of all the bullet holes present on planes that came back from their missions. After going through thousands of pages of documents, detailing the highest frequency areas of damage, the leader of the group, Abraham Wald, made a startling recommendation. "Put the armor where the bullets aren't". To the untrained mind this logic doesn't appear to make a lot of sense. Wouldn't you want to reinforce the areas with the greatest amount of damage? In short, no. 

The reasoning behind this lies in selection bias. You see, the data provided to Wald and the SRG only encompassed the planes that came back safely from their missions. What you don't see in the data are the planes that failed to make it back. Of the planes that made it back there was minimal damage to the engines and fuel systems. There are two possible explanations of the data. Either those areas with minimal damage are somehow hit less frequently by enemy fire or the planes that had damage to those systems were among the casualties that didn't return.

The truth is the areas with the most recorded amounts of damage actually represented systems that are resilient enough to survive without armor. Because planes with damaged engines and fuel systems didn't make it back, the data didn't represent their damage. This revelation was put to use by the allied forces saving countless numbers of planes and lives for the remainder of the war. 

Everybody makes assumptions. It can be very easy to let your assumptions lead you down the wrong path. When determining solutions it's just as important to see what isn't represented in the data as what is represented. Remember this line of thinking the next time you're trying to solve a complex problem. 

_ _ _

I hope you enjoyed this post on data analysis and problem solving. If you enjoyed this post be sure to like it and click follow at the top of the page. Leave me a comment if there's anything you'd like to add. You may also enjoy some of my other posts on LinkedIn, including; Leave Your Bank For A Credit Union, The Best First Job or The Persistent Underutilization of Human Capital. You can find all of my posts here.

Nick Reed Smith is an Award Winning Quality Engineering professional who served in the US Navy Submarine Corps for 4 years. He lives and works in Cincinnati, OH with his wife Meena Smith.