It's not the data you collect, it's how you interpret that data that matters.
During WWII, when planes returned with battle damage they were repaired and returned to service. Because so many of the returning planes had damage in the wings and fuselage, there was a suggestion made to add armor to the areas that were showing damage.
That plan might have been carried out except for Abraham Wald. Dr. Wald was a statistician. He looked at the battle damage reports on returning aircraft and noticed the obvious. These were returning planes. That meant that the damage they had received was not enough to prevent them from flying.
Since he understood that aerial combat was never precise enough to hit another aircraft in an exact spot, but was simply shooting and getting hits wherever they might, he was only looking at part of the data. The rest of the data was smashed into the dirt of France and Germany. The planes that did not return held the real information.
Since those planes were unavailable for inspection, he reversed the thinking. Of the planes that had returned with battle damage, where had they not been hit? In the cockpit, engine, and tail. If additional armor plate was going to be effective, that's where it should be installed.
His logic carried the day, the Army Air Corps added armor as he suggested. If you can do graduate level statistics, here's the math behind it.