Finding Sample Size - Data Type is Attribute

How can I build sample plan when I have
only 2 inputs:
The probability of acceptance is 80%
The confidence level is 90%
how can I find the need sample size out of the above data?

The data type is attributes.

Please your assistance
 
Last edited:
Re: Finding sample size

Thank you for your assistance!

All that information that you gave
Me is not new to me, it doesn't help me to solve
the question
The lot size is 100,000 units

Can you help figure out the answer?
 
W

wmarhel

Re: Finding sample size

If memory serves me correctly (it is Monday morning), than the Mil-Std sampling plans were designed to accept product 95% of the time if the lot is at or better than the designated AQL. AQL generally defined as the percent defective (# defects per 100 units x 100%). The results of the sampling plans (pass or fail) could be stated with a confidence level of 95%.

Wayne
 
Re: Finding sample size

You are absolutely correct,

Since I am looking for sample size not based on AQL=95%
but based on the “starting data” of:
Probability of acceptance is 80%
and confidence level is 90%
I can not use the regular table of Mil Std 105.

I assume that Table "X" in this Mil Std can assist but I do not how to
Corporate the “starting data” into real numbers of sample size.
 

Tim Folkerts

Trusted Information Resource
Nissim Shaked,

Perhaps you could clarify what you are after a bit more, because I'm not quite understanding your question.

For sampling, the typical parameters are for the parts include:
  • AQL (the defect rate that is considered acceptable)
  • alpha (the odds of rejecting an acceptable lot)
  • LTPD ("lot tolerance percent defective = defect rate that is considered unacceptable)
  • beta (the odds of accepting an unacceptable lot)
  • Lot Size
Typical parameters for the sampling plan include:
  • N = sample size
  • Ac (or c) = max number of defects that will pass
  • Re = min number of defects that will fail
Presumably your "probabilty of acceptance" is 1-alpha but I'm not sure on the "confidence level". The parameters above are obviously interconnected; once most of hem have be specified, the rest can be calculated.


Tim F


P.S. It turns out that the lot size is not very important. As long as [lot size] >> [sample size] (which is usually the case for sampling plans) then lot size doesn't matter.


P.P.S As Wayne said, alpha is about 95% for "normal" inspection using MIL-STD-105E (same as ASQZ1.4), although it can vary from about 90% to about 99% depending on the specific plan.
 
Dear Tim,

Thank you again for your time and effort to help me.
I am initiating a sample plan for verification tests based on risk analysis results for a new product.
My risk analysis is connected to sample plan by two parameters:
1) Probability of acceptance
2) Confidence level
We have 4 levels of sample size requirements:
For example:
What should be the sample size if we know:
a) That probability of acceptance is 95%
b) And the confidence level is 90%
The data type is attribute.
I assume that table X-A-1 in Mil-Std 105 can help to find the answer, but I am not familiar with the connection of the 2 parameters with a sample size.
I know that there is a way to do it , but again I don’t know how to do it.

Thank you in advance
Nissim
:confused:
 
Top Bottom