Lets assume I have an effect, that affects/contaminates every lot (of production, typical 15k) different, something between 0,0% and 6,0%. Now I have changed something and want to test whether this change my contamination or not. I have produced four testlots with 1k and tested them. I found no affected members in those four lots.
Now I have to estimate an upper bound for contamination. How could that be done? I need something like "We can be pretty sure (alpha = 5%) that containment is lower then 0.5%".