Determining Sample Size for Attribute Data

N

nwillnow

Hey everyone. I'm currently trying to prove that a system I have in place is reliable to the severity level required and I'm in need of sampling plan information. I have a standard operation procedure which tells me that for 95% confidence and 95% reliability (1.0 AQL) that I need 59 samples for c=0, and 95 for c=1, but no information for c=2,3,4,...

All it tells me about the plan is that "The attribute data sample
sizes are based on the binomial distribution for the corresponding reliability and confidence level requirements." I've tried using Statgraphics and MiniTab to help calculate what the c=2 sample would be, but I can't figure out how.

Is this a common standard I'm looking at? My problem is that I'm working backwards. I'm trying to use pass/fail data I have collected over a number of weeks to show that the number of failures I'm seeing per week is acceptable, as opposed to structuring a validation around a sample size to test.

I've looked at the ANSI/ASQC Z1.4 standard, but that seems entirely different. Any suggestions?
 

Tim Folkerts

Trusted Information Resource
A while back I posted a spreadsheet for calculating sampling plans and it can determine the numbers you are looking for:
Sampling Spreadsheet to help determine the sample size for sampling plans

For 95% reliability, enter 5% in the "Unacceptable Quality Level" cell
For 95% certainty, enter 5% in the "Beta Risk" cell

Then adjust the sample size cell (or the slider) and watch the "RE" cell (the reject number). Once it changes, then you have the smallest sample needed to acheive your goals. Keep increasing the sample size and you will reach Re=1, Re=2, ... I checked a few and they agree with the numbers posted already.

Obviously, you could change the "Unacceptable Quality Level" and the "Beta Risk" for other levels. You can also "reverse engineer" the formulas to see how I got the results.


Tim
 
N

nwillnow

Sound. Cheers for that lads! Never would have found that on me own.
 
Top Bottom