Hi,
Yes, I have 10 failures in 500 parts, but those devices don’t see the same stress and therefore don’t have the same failure probabilities. I’m trying to estimate the environmental stress distribution from the number of failures but it would be too restricting to assume that the stress is the same on all of the devices. The last bit with the normal distribution is just a crutch because I didn’t know how else to compare predicted nfailures with observed nfailures. It would also be good if I could get rid of that, but I’m not sure if that will fix the sampling problem.