In trying to decide what to do about school shootings, especially mass shootings, are we making a fundamental error by considering only the schools that suffered them and the men who committed them? Or should we base our decisions mainly by examining schools that have not suffered them?
In my prior post on why we are probably alone in the universe ("Where is everybody? They're dead"), I included what the analysis of combat-damage assessment of World War II bombers had to do with the existence of life on other worlds. What might it have to do with school shootings? Let's see.
The Atlantic wrote of World War 2 bomber crews who did pattern analysis of bullet holes from enemy fighters attacks. They thought if there was a pattern, then they could lessen the number of bombers shot down by increasing the armor in the hit sections of the bomber.
But the Hungarian-born mathematician Abraham Wald, and his colleagues at the Statistical Research Group at Columbia University, had a novel, if counterintuitive, prescription. Don’t protect the planes where they were taking the most damage, Wald said. Armor the planes where there were no bullet holes at all.
“You put armor where there are no holes, because the planes that got shot there didn’t return to the home base,” says Anders Sandberg, a senior research fellow at University of Oxford’s Future of Humanity Institute. “They crashed.”
The holes didn’t show where returning planes were likely to get hit, but only what it was possible for later observers to see. This is known as an observer selection effect, and the same sort of bias might apply not only to perforated planes, but to whole worlds as well.
Observer selection effect (OSE henceforth) means that we are able to observe something only because we survived the causes. It is not the same as observer bias, in which our interpretations of what we observe is shaped by preconceptions. No, OSE means that we can observe at all because we are observing only non-negative outcomes.
I asked my chemical-engineer daughter about the shot bomber problem and she immediately said that the bombers that did not return were of greater assessment value than the ones who did. She explained that OSE is a well-understood hazard in science and engineering and was a topic of discussion in her statistics and probability courses.
And so to schools: Are we looking in the wrong places for indicators of what to do? Maybe the characteristics of schools and their students who have not been subjected to the violence could be more informative than the ones who have.
Or maybe there are simply too few school massacres (fortunately) to do an OSE analysis. I do not know, I am neither an engineer nor mathematician. But if so, I wonder whether that might mean we are doomed to know too little to do anything effective until, God forbid, there have been a large number more such horrors.
“You put armor where there are no holes, because the planes that got shot there didn’t return to the home base,” says Anders Sandberg, a senior research fellow at University of Oxford’s Future of Humanity Institute. “They crashed.”
The holes didn’t show where returning planes were likely to get hit, but only what it was possible for later observers to see. This is known as an observer selection effect, and the same sort of bias might apply not only to perforated planes, but to whole worlds as well.
