Survivorship bias

I encountered this concept in a video I watched on the Bebar Bilim channel by Cengiz Çalışkan, whom we also watch on 12punto's YouTube channel in the 12’de Bilim program, and I wanted to share and interpret this topic with you here.

There is a great story told in the video as well: “During World War II, America was struggling to establish air superiority and was suffering serious losses. Many planes that went on missions did not return. Both very expensive planes were being lost and very valuable pilots. Scientists rolled up their sleeves to solve this. Experts wanted to reinforce the planes and add steel plating to certain sections, but they did not want the plane to be too heavy. Therefore, they examined the planes that returned from the war to determine where the planes needed reinforcement, looking at where they were hit by bullets the most. According to this, the bullets were generally concentrated on the rear part and the fuel section of the planes. There were no bullet holes in the engine section. Concluding that the planes were generally hit in these areas, they naturally decided to reinforce these areas immediately. However, Hungarian statistician and mathematician Abraham Wald said the exact opposite. It does not matter where the returning planes were hit. The planes you are looking at are the ones that were able to return; the damage they received did not cause them to crash. If you had the opportunity to look at the ones that crashed, you would most likely see that most of them were hit in the engine section. Therefore, he says, you should reinforce the areas where the returning planes were statistically hit the least, especially the engine section. They did so, and there was a significant increase in the number of returning planes, and America began to establish air superiority.”

As this example shows, examining only the examples that succeed or survive causes us to ignore the data of those that fail or do not survive. This leads us to reach conclusions that are not entirely accurate or are based on incomplete data.

The impact of this error on communication studies and social research is also quite significant. For example, in the analysis of media content, focusing only on content that is popular and reaches large audiences may help us understand why some content is successful, but it also causes us to ignore why unsuccessful content fails. Therefore, it can lead to incomplete or misleading media strategies.

In summary, recognizing and avoiding Survivorship Bias helps us make more holistic and accurate analyses. It enables us to develop more effective strategies and reach more accurate results. In this context, it is of great importance in communication studies and social research to consider all aspects of the analyzed data and not to ignore missing data. In short, we must always take the other side of the coin into account and focus not only on what is visible but also on what is not visible at first glance.