How good is a sample size of 1? About 30 years ago, an engineer walked into my office asking: "How good is a sample size of 1?" Our company had decided to export one of our vehicles lines which was already sold in the United States to a South American country. Because it was being sold in the United States and being shipped with no design changes, it was assumed that no additional testing was necessary. It turns out that there was dirt in the motor oil in that country, and every single engine in the vehicles exported had to be replaced (huge warranty bill). What is your estimate of the chances that testing 1 vehicle would have uncovered this issue, assuming more than 500 were exported