iSixSigma

Audit Sampling

Six Sigma – iSixSigma Forums Old Forums General Audit Sampling

Viewing 6 posts - 1 through 6 (of 6 total)
  • Author
    Posts
  • #36923

    Ronald
    Participant

    Hi there!  Just want to ask on what sampling plan is best to use when conducting process audits?  Like for 400 workers, how many of them will be the suitable sample to represent the performance of the whole population?
    Thanks !

    0
    #107804

    Rajesh Narasimhan
    Participant

    Lee
       I will recommend Stratified Random Sampling.  Grouping the homeogenous group in one strata and sampling it from there on.
     
    Thanks

    0
    #107812

    Ronald
    Participant

    Thanks for the reply Rajesh !
    I was thinking of the same but I don’t know if it has a statistical basis. If I stratify the group, how many of them would I get for audit? Say, out of 400 people and from there I was able to form 3 groups. One group composed of 100 people.  How many samples will I get to say that I have covered them all in terms of sampling?
    Can you further enlighten me on this?
    Warm regards,
    Lee Chu
     
     

    0
    #107827

    Dillon
    Participant

    Lee,
    Usually the sampling during a process audit isn’t trying to be “statistically valid.” Most process auditors will audit 3 to 5 people within each unique group of people, and assess from there whether the process is defined and in control. It’s quite possible that noncomplaince will be overlooked, but it’s a trade-off.
    —Doug
     
     

    0
    #107829

    Kulkarni
    Member

    Hi,
    I have also used stratified random sampling plan in one of inspection process, but not sure about the sample % to be picked. Further to this If you have any idea about Acceptance Quality Level (AQL) concept and based on this number of defects allowed in a sample, that would be of great help.
    Vikrant
     
     
     

    0
    #108026

    Rajesh Narasimhan
    Participant

    Hi
    Let us say you have formed 4 groups each with 100 members.  Within the group all the members should be homogenous, in sense when you define your group let say that you did the grouping based on designation, all the people who are having the same designation must be in one group.  Further within the group you can divide those 100 into manageable subgroups.  For example, within 100 you can subdivide the group based on age, sex , experience or whatever factors that you think good for your analysis.  After subdividing it, from the group 100, I would pick a sample of 1/5 th of 100 which is 20.  Let us say you have 4 subgroups then from each group I will pick 5 of them for my analysis.  Hope I have clarified your point.  If you really want to have more info, you can go thru the book call Sampling Techniques by Ranjith Singh Aurora.  Otherwise let me know what is your problem definition, so that I can help you out.
    Thanks
    Rajesh Narasimhan
     
     

    0
Viewing 6 posts - 1 through 6 (of 6 total)

The forum ‘General’ is closed to new topics and replies.