Читать книгу Paint Analysis - Roger Dietrich - Страница 30

How poor can quality be consistent with the finding that a random sample contains zero defects? The first example calculated above implies the knowledge of a defect rate. But often you do not know it. But you want to be sure that your production lot has zero defects without testing the whole population. Therefore, the question is:

If you test a certain number of samples on a random basis, how sure can you be that the whole population is ok?

Let us assume that 100 samples are selected randomly from a production line and zero defects have been found. So, n is 100. Following the rule of 3 σ the upper risk level is 3/100 or 0.03 with a confidence level of 95 %. This means that the tested lot can contain up to 3 % defects even if zero defects have been found in a random sample of 100!

Note: If nothing goes wrong, it is not necessarily all right!