How do I draw a OC (Operating Curve), given the sampling plan, lot size, sample size, and acceptance no?
For eg: Lot size: 2000, Sample size: 200, and acceptance no. is 4, how would I draw the OC Curve?
You’re gonna get some specific replies from more seasoned statisticians. However – why in the world would you want to draw the curve in the first place? Sounds like a bit of self torture to me. Personally, if I never see an OC curve again, it will be too soon.
Get a copy of Grant and Levenworths book “Quality Control Handbook” (I think – I am on the road and don’t have it with me). They tell you how to do it. Back half of the book.
If you are still doing acceptance sampling and it is effective then your quality level is so poor you are probably better off fixing something than drawing an OC curve.
In theory the acceptance number of 4 (rather than 0) will create a curve that represents theoretically what the curve should look like. The message it sends makes you appear pretty cavalier about what you are shipping. You find defects, but not to many and you still ship it? That is about 30 years behind the times.
Just my opinion.
The OC curve has the probability of accepting the lot on the y-axis and the actual proportion defective on the x-axis (just to be clear, 1% defective would be 0.01). The following Excel formula will generate the probability of acceptance: BINOMDIST(c, n, p, TRUE) where c = accept number, n = sample size, p = actual proportion defective, and TRUE is a cumulative flag. To draw an OC curve (or even to look one up in a book of curves), you need to know c and n. Put values from 0 to 1 in a column to use as p, and put the formula in a second column. Have Excel draw a curve with p as the x-axis and the formula results as the y-axis.
I have found one good use for drawing OC curves – it tends to scare the hell out of people when they can see the probability of accepting a lot of product that is (for example) really 15% defective, even if the AQL is 1%.
The forum ‘General’ is closed to new topics and replies.