My daily business
If need be, I select and test a sample on the basis of probable error and then write something like "14 out of a risk-based sample of 20 items were deficient." That the addressee of my report will think about an error rate of 70 per cent - well, that's part of the game.
Of course, to the best of my knowledge it might just as well be a total of 14 errors out of a population of umpteen thousands. But based on the approach it is simply not possible to draw any conclusions on the population.
Biting the hand that feeds IT
