Chapter 15. Coda: Understanding data with graphs
But in the end, the most important thing is to draw the right conclusions from the available evidence.
During World War II, a program was launched to explore the possibility of equipping fighter planes with additional armor as protection against ground-based anti-aircraft fire.[1] To determine where on the airplane to place the armor, fighter planes returning from combat missions were investigated for bullet holes. In figure 15.1, the areas where bullet holes were found are shown shaded; areas without bullet holes are left white.
1 The inspiration to this story stems from the book Graphic Discovery by Howard Wainer, Princeton University Press (2005).
Figure 15.1. Schematic outline of a fighter airplane. Areas where bullet holes were found on machines returning from combat missions are shaded. Where would you recommend additional armor to be placed?
Given this evidence, where would you put additional armor?
The not-so-obvious obvious answer is to add the armor in those areas where no bullet holes were found. Why? Because airplanes are subject to hits everywhere, but if the hits strike in the white areas in figure 15.1, the airplane doesn’t come back from its mission. (Statisticians speak of survivorship bias.) Therefore, those are the most vital areas of the machine and should receive the best possible protection.