How do you calculate an AQL Confidence Level?

Q

QApoultry

Help with a question? How do you figure Confidence level?
I had a inspector/auditor want a lot of product 12,000 lbs inspected at a 95% confidence level?

Obviously he wasn't too skilled on AQL and Statistics but neither am I.

Based on my ANSI slide rule that would be Code letter M or 315 pounds
base on inspection level II.
No can someone help me how far am I off and I guess I am getting AQL and confidence levels confused. Any assistance would be appreciated.
 
D

Duke Okes

Typically we don't calculate an AQL, we set one then select a sampling plan that will achieve it.
 

somashekar

Leader
Admin
Typically we don't calculate an AQL, we set one then select a sampling plan that will achieve it.
The confidence level tells you how sure you can be. It is expressed as a percentage and represents how often the true percentage of the population who would pick an answer lies within the confidence interval. The 95% confidence level means you can be 95% certain; the 99% confidence level means you can be 99% certain.
So you have a sampling plan for various AQL determined seperately for a 95% confidence level and 99% confidence level. You determine the confidence level you want to apply.
 

Tim Folkerts

Trusted Information Resource
I see several difficulties with your situation.


  • AQL is typically applied to discrete objects, not to bulk materials (which is what I assume you have here). So AQL would be a strange choice to begin with in your case. It could be done -- basically assuming that you have 12000 individual lots of 1 pound.
  • If you can measure a "variable" (like the actual density of the sample) instead of an "attribute" (like "yes or no: is the density under 2.3 g.cm^3), then you can use a variable sampling plan, which typically requires a MUCH smaller sample to achieve the same level of confidence.
  • AQL plans provide pretty much the oppose of the sort of confidence you want. A typical Level II normal inspection tells you that you are "NOT 95% confident the lot is bad" so you accept the lot, even though bad lots could get by quite often. You presumable want to be "95% confident the lot is NOT bad" so you can accept the lot with confidence. These are two very different things.
  • You need to state WHAT you want to be 95% confident of. For example "95% confident that there is no more than 1% foreign material," which would lend itself to variable testing. You could also do something like "95% confident that 1 lb samples would pass inspection 99% of the time," which would lend itself to attribute sampling.
  • Furthermore, the choice to use pounds as your unit is completely arbitrary. You could just as well call it 5500 kg or 6 tons. Those choices would obviously change the sample sizes and the inspection costs. Generally I would suggest finding the smallest unit that would give reliable results. Maybe you can test just an ounce and get a good result. maybe you need to test a pound or 10 pounds to get a representative sample.
I'm sure others will weigh in with more advice.


Tim F
 
G

George Weiss

Is the auditor attempting to guide you into ISO/IEC-17025:2005 testing compliance?
When I hear 95% I think of 17025 coverage factor of 2, (k2), and 99% being a coverage factor of 3, (k3).
Determine what is to be tested for, and sample test for these items with a k2 or k3 coverage as in 17025
 

Bev D

Heretical Statistician
Leader
Super Moderator
yes - your auditor's request is non-sensical as they have incompletely specified their request for inspection.
 
J

Jim G101

If I understand correctly then, you would have to use each Sample Size Code Letter on its own merit, correct? Example:

I have a team of 8 personnel doing a task to the same work instruction/ performance standard. I want to do a sample size of those 8 personnel. As an industry norm, I have decided to use GIL 2, w/ an AQL of 2.5. GIL 2 says use Sample Size Code Letter A. Using Single, Normal chart, actually evaluate 2 personnel. 2.5 AQL says accept the performance of the team if both operators are complying with requirements and reject (fail the team) if one of those operators is not complying with instructions (fails). For management, I can tell them we have a 90% Confidence Level that evaluations will only miss 2.11 non conformances out of every 100 observations (instruction requirements/ checklist items) or a 90% Confidence Level that we will accept team performance 97.91% of the time. (No 2.5 AQL in tables for A and B, go to Table X-C-1, 90% Pa (loosely a 90% Confidence Level). Is this correct?
 
Top Bottom